Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 108
Encontrada(s) 2 página(s)

e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publicación Contribución de congreso (ConferenceOutput)

NUMA impact on network storage protocolsover high-speed raw Ethernet

  • González-Férez, Pilar
  • Bilas, Angelos
Proceedings of: Second International Workshop on Sustainable Ultrascale Computing Systems (NESUS 2015). Krakow (Poland), September 10-11, 2015., Current storage trends dictate placing fast storage devices in all servers and using them as a single distributed storage system. In this converged model where storage and compute resources co-exist in the same server, the role of the network is becoming more important: network overhead is becoming a main imitation to improving storage performance. In our previous work we have designed Tyche, a network protocol for converged storage that bundles multiple 10GigE links transparently and reduces protocol overheads over raw Ethernet without hardware support. However, current technology trends and server consolidation dictates building servers with large amounts of resources (CPU, memory, network, storage). Such servers need to employ Non-Uniform Memory Architectures (NUMA) to scale memory performance. NUMA introduces significant problems with the placement of data and buffers at all software levels. In this paper, we first use Tyche to examine the performance implications of NUMA servers on end-to-end network storage performance. Our results show that NUMA effects have significant negative impact and can reduce throughput by almost 2x on servers with as few as 8 cores (16 hyper-threads). Then, we propose extensions to network protocols that can mitigate this impact. We use information about the location of data, cores, and NICs to properly align data transfers and minimize the impact of NUMA servers. Our design almost entirely eliminates NUMA effects by encapsulating all protocol structures to a “channel” concept and then carefully mapping channels and their resources to NICs and NUMA nodes., We thankfully acknowledge the support of the European Commission under the 7th Framework Programs through the NanoStreams (FP7-ICT-610509) project, the HiPEAC3 (FP7-ICT-287759) Network of Excellence, and the COST programme Action IC1305, ’Network for Sustainable Ultrascale Computing (NESUS)’.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Power-aware scheduling with effective task migration for real-time multicore embedded systems

  • March Cabrelles, José Luis
  • Duato Marín, José Francisco
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Hassan Mohamed, Houcine|||0000-0002-2038-0587
A major design issue in embedded systems is reducing the power consumption because batteries have a limited energy budget. For this purpose, several techniques such as dynamic voltage and frequency scaling (DVFS) or task migration are being used. DVFS allows reducing power by selecting the optimal voltage supply, whereas task migration achieves this effect by balancing the workload among cores. This paper focuses on power-aware scheduling allowing task migration to reduce energy consumption in multicore embedded systems implementing DVFS capabilities. To address energy savings, the devised schedulers follow two main rules: migrations are allowed at specific points of time and only one task is allowed to migrate each time. Two algorithms have been proposed working under real-time constraints. The simpler algorithm, namely, single option migration (SOM) only checks just one target core before performing a migration. In contrast, the multiple option migration (MOM) searches the optimal target core. In general, the MOM algorithm achieves better energy savings than the SOM algorithm, although differences are wider for a reduced number of cores and frequency/voltage levels. Moreover, the MOM algorithm reduces energy consumption as much as 40% over the worst fit algorithm., This work was supported by the Spanish MICINN, Consolider Programme and Plan E funds, as well as European Commission FEDER funds, under Grants CSD2006-00046 and TIN2009-14475-C04-01.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Exploiting Reuse Information to Reduce Refresh Energy in On-Chip eDRAM Caches

  • Valero Bresó, Alejandro
  • Duato Marín, José Francisco
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
This work introduces a novel refresh mechanism that leverages reuse information to decide which blocks should be refreshed in an energy-aware eDRAM last-level cache. Experimental results show that, compared to a conventional eDRAM cache, the energy-aware approach achieves refresh energy savings up to 71%, while the reduction on the overall dynamic energy is by 65% with negligible performance losses., This work was supported by the Spanish Ministerio de Economía y Competitividad (MINECO) and Plan E funds, under Grants TIN-2009-14475-C04-01 and TIN2012-38341-C04-01.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

L1-Bandwidth Aware Thread Allocation in Multicore SMT Processors

  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Duato Marín, José Francisco
Improving the utilization of shared resources is a key issue to increase performance in SMT processors. Recent work has focused on resource sharing policies to enhance the processor performance, but their proposals mainly concentrate on novel hardware mechanisms that adapt to the dynamic resource requirements of the running threads. This work addresses the L1 cache bandwidth problem in SMT processors experimentally on real hardware. Unlike previous work, this paper concentrates on thread allocation, by selecting the proper pair of co-runners to be launched to the same core. The relation between L1 bandwidth requirements of each benchmark and its performance (IPC) is analyzed. We found that for individual benchmarks, performance is strongly connected to L1 bandwidth consumption, and this observation remains valid when several co-runners are launched to the same SMT core. Based on these findings we propose two L1 bandwidth aware thread to core (t2c) allocation policies, namely Static and Dynamic t2c allocation, respectively. The aim of these policies is to properly balance L1 bandwidth requirements of the running threads among the processor cores. Experiments on a Xeon E5645 processor show that the proposed policies significantly improve the performance of the Linux OS kernel regardless the number of cores considered., This work was supported by the Spanish Ministerio de Econom´ıa y Competitividad (MINECO) and by FEDER funds under Grant TIN2012-38341-C04-01; and by Programa de Apoyo a la Investigacion y Desarrollo (PAID-05-12) of the ´ Universitat Politecnica de Val ` encia under Grant SP20120748
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Introduction to the Special Section on On-Chip and Off-Chip Network Architectures

  • Flich Cardo, José
  • Palesi, Maurizio
Interconnection networks are pervasive in most current systems, spanning large supercomputer installations to multiprocessor chips and embedded systems. Every year, the number of processing elements to interconnect increases both at the cluster level and at the chip level, so the performance of the interconnect becomes a predominant factor.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

On Microarchitectural Mechanisms for Cache Wearout Reduction

  • Valero Bresó, Alejandro
  • Miralaei, Negar
  • Jones, Timothy M.
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] Hot carrier injection (HCI) and bias temperature instability (BTI) are two of the main deleterious effects that increase a transistor's threshold voltage over the lifetime of a microprocessor. This voltage degradation causes slower transistor switching and eventually can result in faulty operation. HCI manifests itself when transistors switch from logic "0" to "1" and vice versa, whereas BTI is the result of a transistor maintaining the same logic value for an extended period of time. These failure mechanisms are especially acute in those transistors used to implement the SRAM cells of first-level (L1) caches, which are frequently accessed, so they are critical to performance, and they are continuously aging. This paper focuses on microarchitectural solutions to reduce transistor aging effects induced by both HCI and BTI in the data array of L1 data caches. First, we show that the majority of cell flips are concentrated in a small number of specific bits within each data word. In addition, we also build upon the previous studies, showing that logic "0" is the most frequently written value in a cache by identifying which cells hold a given logic value for a significant amount of time. Based on these observations, this paper introduces a number of architectural techniques that spread the number of flips evenly across memory cells and reduce the amount of time that logic "0" values are stored in the cells by switching OFF specific data bytes. Experimental results show that the threshold voltage degradation savings range from 21.8% to 44.3% depending on the application., This work was supported in part by the Spanish Ministerio de Economia y Competitividad within the Plan E Funds under Grant TIN2015-66972-C5-1-R, in part by the HiPEAC Collaboration Grant funded by the FP7 HiPEAC Network of Excellence under Grant 287759, and in part by the Engineering and Physical Sciences Research Council under Grant EP/K026399/1 and Grant EP/J016284/1.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Exploring the interoperability of remote GPGPU virtualization using rCUDA and directive-based programming models

  • Castelló, Adrián|||0000-0002-8576-8451
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Peña Monferrer, Antonio José
  • Mayo Gual, Rafael
  • Planas,Judit
  • Balaji, Pavan
[EN] Directive-based programming models, such as OpenMP, OpenACC, and OmpSs, enable users to accelerate applications by using coprocessors with little effort. These devices offer significant computing power, but their use can introduce two problems: an increase in the total cost of ownership and their underutilization because not all codes match their architecture. Remote accelerator virtualization frameworks address those problems. In particular, rCUDA provides transparent access to any graphic processor unit installed in a cluster, reducing the number of accelerators and increasing their utilization ratio. Joining these two technologies, directive-based programming models and rCUDA, is thus highly appealing. In this work, we study the integration of OmpSs and OpenACC with rCUDA, describing and analyzing several applications over three different hardware configurations that include two InfiniBand interconnections and three NVIDIA accelerators. Our evaluation reveals favorable performance results, showing low overhead and similar scaling factors when using remote accelerators instead of local devices., The researchers from the Universitat Jaume I de Castello were supported by Universitat Jaume I research project (P11B2013-21), project TIN2014-53495-R, a Generalitat Valenciana grant and FEDER. The researcher from the Barcelona Supercomputing Center (BSC-CNS) Lausanne was supported by the European Commission (HiPEAC-3 Network of Excellence, FP7-ICT 287759), Intel-BSC Exascale Lab collaboration, IBM/BSC Exascale Initiative collaboration agreement, Computacion de Altas Prestaciones VI (TIN2012-34557) and the Generalitat de Catalunya (2014-SGR-1051). This work was partially supported by the U.S. Dept. of Energy, Office of Science, Office of Advanced Scientific Computing Research (SC-21), under contract DE-AC02-06CH11357. The initial version of rCUDA was jointly developed by Universitat Politecnica de Valencia (UPV) and Universitat Jaume I de Castellon (UJI) until year 2010. This initial development was later split into two branches. Part of the UPV version was used in this paper. The development of the UPV branch was supported by Generalitat Valenciana under Grants PROMETEO 2008/060 and Prometeo II 2013/009. We gratefully acknowledge the computing resources provided and operated by the Joint Laboratory for System Evaluation (JLSE) at Argonne National Laboratory.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

The impact of User-Browser Interaction on web performance

  • Peña Ortiz, Raúl
  • Gil Salinas, José Antonio
  • Pont Sanjuan, Ana
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
The user interaction with the current web contents is a major concern when defining web workloads in order to precisely estimate system performance. However, the intrinsic diffi- culty to represent this behavior in a workload model leads many research works to still use workloads non representative of the current web navigations. In contrast, in previous works we demonstrated that the use of an accurate workload model that considers user’s dynamism when navigating the web clearly affects system performance metrics. In this paper we analyze, for the first time, the effect of considering the User-Browser Interaction (UBI) as a part of user’s dynamic behavior on web workload characterization in performance studies. To this end, we evaluate a typical e-commerce scenario and compare the obtained results for different UBI behaviors, such as the use of the back button and parallel browsing originated by using browser tabs or opening new windows when surfing a website., This work has been partially supported by the Spanish Ministry of Science and Innovation under grant TIN-2009-08201.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Smartphone-based industrial informatics projects and laboratories

  • Hassan Mohamed, Houcine|||0000-0002-2038-0587
  • Martínez-Rubio, Juan-Miguel|||0000-0002-9748-0332
  • Perles, Angel|||0000-0002-5782-4284
  • Capella Hernández, Juan Vicente|||0000-0003-3879-9826
  • Domínguez Montagud, Carlos Pascual
  • Albaladejo Meroño, José
The use of IT technologies plays an important role in the training of future engineers. In this paper, smartphones and multimedia technologies are proposed as an innovative way to tackle the formation of students, at different levels, in the Industrial Informatics (II) subject of the Industrial Electronics Engineering (IEE) degree. II instructs future Engineers in the design of IT systems to control industrial processes. In the first level, smartphones are used to display a web-based multimedia tool that is implemented to register the lecture explanations regarding the design of II systems, so as it facilitates student to guide him/her self in the learning process. In the second level, the smartphone is proposed as the control system of a medium size industrial process (e.g., water tank). Since II uses a problem-based learning methodology (miniproject) to instruct the design of II systems, for each lecture, laboratory practices are tackled, and the solutions obtained are embedded in the smartphone to control the corresponding part of the miniproject. An application of the Smartphone multimedia tool is presented to show how students interact with the developed system. The successful evaluation of the proposed tools, by more than 900 IEE students during three years, is shown., This work was supported in part by the Universitat Politecnica de Valencia, Valencia, Spain, under Grant 20090513-" Dynamizing the European Convergence Higher Education." Paper no. TII-11-308.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Increasing the effectiveness of directory caches by avoiding the tracking of noncoherent memory blocks

  • Cuesta Sáez, Blas Antonio
  • Ros Bardisa, Alberto
  • Robles Martínez, Antonio
  • Duato Marín, José Francisco
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
A key aspect in the design of efficient multiprocessor systems is the cache coherence protocol. Although directory-based protocols constitute the most scalable approach, the limited size of the directory caches together with the growing size of systems may cause frequent evictions and, consequently, the invalidation of cached blocks, which jeopardizes system performance. Directory caches keep track of every memory block stored in processor caches in order to provide coherent access to the shared memory. However, a significant fraction of the cached memory blocks do not require coherence maintenance (even in parallel applications) because they are either accessed by just one processor or they are never modified. In this paper, we propose to deactivate the coherence protocol for those blocks that do not require coherence. This deactivation means directory caches do not have to keep track of noncoherent blocks, which reduces directory cache occupancy and increases its effectiveness. Since the detection of noncoherent blocks is carried out by the operating system, our proposal only requires minor hardware modifications. Simulation results show that, thanks to our proposal, directory caches can avoid the tracking of about 66 percent (on average) of the blocks accessed by a wide range of applications, thereby improving the efficiency of directory caches. This contributes either to shortening the runtime of parallel applications by 15 percent (on average) while keeping directory cache size or to maintaining performance while using directory caches 16 times smaller., This work was supported by the Spanish MICINN, Consolider Programme and Plan E funds, as well as European Commission FEDER funds, under Grants CSD2006-00046 and TIN2009-14475-C04-01. It was also partly supported by (PROMETEO from Generalitat Valenciana (GVA) under Grant ROMETEO/2008/060). B. Cuesta was with Universitat Politecnica de Valencia while working on this paper.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

A fault-tolerant routing strategy for k-ary n-direct s-indirect topologies based on intermediate nodes

  • Peñaranda Cebrián, Roberto
  • Gran, Ernst Gunnar
  • Skeie, Tor
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • López Rodríguez, Pedro Juan|||0000-0003-4544-955X
[EN] Exascale computing systems are being built with thousands of nodes. The high number of components of these systems significantly increases the probability of failure. A key component for them is the interconnection network. If failures occur in the interconnection network, they may isolate a large fraction of the machine. For this reason, an efficient fault-tolerant mechanism is needed to keep the system interconnected, even in the presence of faults. A recently proposed topology for these large systems is the hybrid k-ary n-direct s-indirect family that provides optimal performance and connectivity at a reduced hardware cost. This paper presents a fault-tolerant routing methodology for the k-ary n-direct s-indirect topology that degrades performance gracefully in presence of faults and tolerates a large number of faults without disabling any healthy computing node. In order to tolerate network failures, the methodology uses a simple mechanism. For any source-destination pair, if necessary, packets are forwarded to the destination node through a set of intermediate nodes (without being ejected from the network) with the aim of circumventing faults. The evaluation results shows that the proposed methodology tolerates a large number of faults. For instance, it is able to tolerate more than 99.5% of fault combinations when there are 10 faults in a 3-D network with 1000 nodes using only 1 intermediate node and more than 99.98% if 2 intermediate nodes are used. Furthermore, the methodology offers a gracious performance degradation. As an example, performance degrades only by 1% for a 2-D network with 1024 nodes and 1% faulty links., This work was supported by the Spanish Ministerio de Economía y Competitividad (MINECO), by FEDER funds under Grant TIN2015-66972-C5-1-R, by Programa de Ayudas de Investigación y Desarrollo (PAID) from Universitat Politècnica de alència and by the financial support of the FP7 HiPEAC Network of Excellence under grant agreement 287759
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

An effective and feasible congestion management technique for high-performance MINs with tag-based distributed routing

  • Escudero-Sahuquillo, Jesús
  • Garcia Garcia, Pedro-Javier
  • Quiles Flor, Francisco Jose
  • Duato Marín, José Francisco
  • Flich Cardo, José
As parallel computing systems increase in size, the interconnection network is becoming a critical subsystem. The current trend in network design is to use as few components as possible to interconnect the end nodes, thereby reducing cost and power consumption. However, this increases the probability of congestion appearing in the network. As congestion may severely degrade network performance, the use of a congestion management mechanism is becoming mandatory in modern interconnects. One of the most cost-effective proposals to deal with the problems derived from congestion situations is the Regional Explicit Congestion Notification (RECN) strategy, based on using special queues to totally isolate the packet flows which contribute to congestion, thereby preventing the Head-of-Line (HoL) blocking effect that these flows may cause to others. Unfortunately, RECN requires the use of source-based routing, thus not being suitable for interconnects with distributed routing, like InfiniBand. Although some RECN-like mechanisms have been proposed for distributed-routing networks, they are not scalable due to the huge amount of control memory that they require in medium-size or large networks. In this paper, we propose Distributed-Routing-Based Congestion Management (DRBCM), a new scalable technique which, following the RECN principles, totally prevents congestion from producing HoL-blocking in multistage interconnection networks (MINs) using tag-based distributed routing. Simulation results indicate that, regardless of network size, DRBCM presents small resource requirements to keep network performance at maximum level even in scenarios of heavy congestion, where it utterly outperforms (with a gain up to 70 percent) current solutions for distributed-routing networks, like the InfiniBand congestion-control mechanism based on injection throttling. Thus, DRBCM is an efficient, cost-effective, and scalable solution for congestion management., This work was jointly supported by the MEC, MICINN (currently MINECO), and European Commission under the projects Consolider Ingenio-2010-CSD2006-00046 and TIN2009-14475-C04, and by the JCCM under projects PCC08-0078 (PhD. grant A08/048) and POII10-0289-3724.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Analyzing web server performance under dynamic user workloads

  • Peña Ortiz, Raúl
  • Gil Salinas, José Antonio
  • Pont Sanjuan, Ana
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
The increasing popularity of web applications has introduced a new paradigm where users are no longer passive web consumers but they become active contributors to the web, specially in the contexts of social networking, blogs, wikis or e-commerce. In this new paradigm, contents and services are even more dynamic, which consequently increases the level of dynamism in user's behavior. Moreover, this trend is expected to rise in the incoming web. This dynamism is a major adversity to define and model representative web workload, in fact, this characteristic is not fully represented in the most of the current web workload generators. This work proves that the web user's dynamic behavior is a crucial point that must be addressed in web performance studies in order to accurately estimate system performance indexes. In this paper, we analyze the effect of using a more realistic dynamic workload on the web performance metrics. To this end, we evaluate a typical e-commerce scenario and compare the results obtained using different levels of dynamic workload instead of traditional workloads. Experimental results show that, when a more dynamic and interactive workload is taken into account, performance indexes can widely differ and noticeably affect the stress borderline on the server. For instance, the processor usage can increase 30% due to dynamism, affecting negatively average response time perceived by users, which can also turn in unwanted effects in marketing and fidelity policies. © 2012 Elsevier B.V. All rights reserved., This work has been partially supported by the Spanish Ministry of Science and Innovation under Grant TIN-2009-08201.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Energy-efficient execution of dense linear algebra algorithms on multi-core processors

  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Dolz Zaragozá, Manuel Francisco
  • Mayo, Rafael
  • Quintana-Ortí, Enrique S.
This paper addresses the efficient exploitation of task-level parallelism, present in many dense linear alge- bra operations, from the point of view of both computa- tional performance and energy consumption. The strategies considered here, referred to as the Slack Reduction Algo- rithm (SRA) and the Race-to-Idle Algorithm (RIA), adjust the operation frequency of the cores during the execution of a collection of tasks (in which many dense linear alge- bra algorithms can be decomposed) with very different ap- proaches to save energy. The procedures are evaluated using an energy-aware simulator, which is in charge of schedul- ing/mapping the execution of these tasks to the cores, lever- aging dynamic frequency voltage scaling featured by current technology. Experiments with this tool and the practical in- tegration of the RIA strategy into a runtime show the energy gains for two versions of the QR factorization., This work was supported by project CICYT TIN2011-23283 and FEDER.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Referrer Graph: A cost-effective algorithm and pruning method for predicting web accesses

  • De La Ossa Perez, Bernardo Antonio
  • Gil Salinas, José Antonio
  • Pont Sanjuan, Ana
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
This paper presents the Referrer Graph (RG) web prediction algorithm and a pruning method for the associated graph as a low-cost solution to predict next web users accesses. RG is aimed at being used in a real web system with prefetching capabilities without degrading its performance. The algorithm learns from users accesses and builds a Markov model. These kinds of algorithms use the sequence of the user accesses to make predictions. Unlike previous Markov model based proposals, the RG algorithm differentiates dependencies in objects of the same page from objects of different pages by using the object URI and the referrer in each request. Although its design permits us to build a simple data structure that is easier to handle and, consequently, needs lower computational cost in comparison with other algorithms, a pruning mechanism has been devised to avoid the continuous growing of this data structure. Results show that, compared with the best prediction algorithms proposed in the open literature, the RG algorithm achieves similar precision values and page latency savings but requiring much less computational and memory resources. Furthermore, when pruning is applied, additional and notable resource consumption savings can be achieved without degrading original performance. In order to reduce further the resource consumption, a mechanism to prune de graph has been devised, which reduces resource consumption of the baseline system without degrading the latency savings. 2013 Elsevier B.V. All rights reserved., This work has been partially supported by Spanish Ministry of Science and Innovation under Grant TIN2009-08201. The authors would also like to thank the technical staff of the School of Computer Science at the Polytechnic University of Valencia for providing us recent and customized trace files logged by their web server.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Enhancing the L1 Data Cache Design to Mitigate HCI

  • Valero Bresó, Alejandro
  • Miralaei, Negar
  • Jones, Timothy M.
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] Over the lifetime of a microprocessor, the Hot Carrier Injection (HCI) phenomenon degrades the threshold voltage, which causes slower transistor switching and eventually results in timing violations and faulty operation. This effect appears when the memory cell contents flip from logic '0' to '1' and vice versa. In caches, the majority of cell flips are concentrated into only a few of the total memory cells that make up each data word. In addition, other researchers have noted that zero is the most commonly-stored data value in a cache, and have taken advantage of this behavior to propose data compression and power reduction techniques. Contrary to these works, we use this information to extend the lifetime of the caches by introducing two microarchitectural techniques that spread and reduce the number of flips across the first-level (L1) data cache cells. Experimental results show that, compared to the conventional approach, the proposed mechanisms reduce the highest cell flip peak up to 65.8 percent, whereas the threshold voltage degradation savings range from 32.0 to 79.9 percent depending on the application., This work has been supported by the Spanish Ministerio de Economia y Competitividad (MINECO), by FEDER funds through Grant TIN2012-38341-C04-01, by the Intel Early Career Faculty Honor Program Award, by a HiPEAC Collaboration Grant funded by the FP7 HiPEAC Network of Excellence under grant agreement 287759, and by the Engineering and Physical Sciences Research Council (EPSRC) through Grants EP/K026399/1 and EP/J016284/1. Additional data related to this publication are available in the data repository at https://www.repository.cam.ac.uk/handle/1810/249006.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Capítulo o parte de un libro (BookPart)

Influence of InfiniBand FDR on the performance of remote GPU virtualization

  • Reaño González, Carlos
  • Mayo Gual, Rafael
  • Duato Marín, José Francisco
  • Peña Monferrer, Antonio José
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Silla, Federico|||0000-0002-6435-1200
The use of GPUs to accelerate general-purpose scientific and engineering applications is mainstream today, but their adoption in current high-performance computing clusters is impaired primarily by acquisition costs and power consumption. Therefore, the benefits of sharing a reduced number of GPUs among all the nodes of a cluster can be remarkable for many applications. This approach, usually referred to as remote GPU virtualization, aims at reducing the number of GPUs present in a cluster, while increasing their utilization rate. The performance of the interconnection network is key to achieving reasonable performance results by means of remote GPU virtualization. To this end, several networking technologies with throughput comparable to that of PCI Express have appeared recently. In this paper we analyze the influence of InfiniBand FDR on the performance of remote GPU virtualization, comparing its impact on a variety of GPU-accelerated applications with other networking technologies, such as Infini-Band QDR and Gigabit Ethernet. Given the severe limitations of freely available remote GPU virtualization solutions, the rCUDA framework is used as the case study for this analysis. Results show that the new FDR interconnect, featuring higher bandwidth than its predecessors, allows the reduction of the overhead of using GPUs remotely, thus making this approach even more appealing., This work was funded by the Generalitat Valenciana under Grant PROMETEOII/2013/009 of the PROMETEO program phase II. This work was supported in part by the U.S. Department of Energy under contract DE-AC02-06CH11357.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Hardware-based generation of independent subtraces of instructions in clustered processors

  • Ubal Tena, Rafael
  • Duato Marín, José Francisco
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • López Rodríguez, Pedro Juan|||0000-0003-4544-955X
Multicore chips are currently dominating the microprocessor market as designs that improve performance and sustain power consumption. However, complex core features must be still considered to provide good performance for existing sequential applications. An effective approach to reduce core complexity without dramatically sacrificing performance is to distribute critical processor structures by using clustered microarchitectures. In these designs, communication latency among clusters is a critical performance bottleneck, and a good steering algorithm is required to reduce intercluster communication. In this paper, we propose a new energy-efficient microarchitectural approach that reduces intercluster communication by detecting and generating independent chains of instructions, referred to as subtraces, from the execution of sequential programs. The devised mechanism has been modeled on an x86-based trace-cache processor, where subtraces are built in the fill unit, stored in a trace cache, and individually steered to different clusters. Experimental results show that the proposal reaches performance speedups around 7 and 15 percent for point-to-point and bus-based interconnects, respectively, while achieving energy savings of up to 12 percent., This work was supported by the Spanish MICINN, Consolider Programme, and Plan E funds, as well as European Commission FEDER funds, under Grants CSD2006-00046 and TIN2009-14475-C04-01.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Built-in fast gather control network for efficient support of coherence protocols

  • Lodde, Mario
  • Roca Pérez, Antoni
  • Flich Cardo, José
[EN] Future chip multiprocessors will include hundreds of cores organised in a tile-based design pattern. These systems commonly employ a shared memory programming model, thus needing a coherence protocol to keep data consistent on the various levels of the cache hierarchy. Usually an invalidation-based protocol is used, where shared copies are invalidated before a write operation. In this study, the authors propose a NoC re-organisation in which a small and fast dedicated control network is used to transmit acknowledgement messages related to the invalidation process, thus relieving the NoC from a considerable percentage of traffic. The dedicated control network is evaluated both with full map directories and with a broadcast-based protocol (Hammer). Experimental evaluation shows significant gains in performance. With a low area overhead (<2.5%), the control network reduces NoC traffic and miss latency, thus reducing execution time up to 16%. Simulation results show a reduction of network traffic up to 80% and a reduction of store and load miss latency up to 70 and 40%, respectively., This work has been supported by the VIRTICAL project (grant agreement n 288574) which is funded by the European Commission within the Research Programme FP7.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

PS-Cache: an energy-efficient cache design for chip multiprocessors

  • Valls, Joan J
  • Ros Bardisa, Alberto
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
Power consumption has become a major design concern in current high-performance chip multiprocessors, and this problem exacerbates with the number of core counts. A significant fraction of the total power budget is often consumed by on-chip caches, thus important research has focused on reducing energy consumption in these structures. To enhance performance, on-chip caches are being deployed with a high associativity degree. Consequently, accessing concurrently all the ways in the cache set is costly in terms of energy. This paper presents the PS-Cache architecture, an energy-efficient cache design that reduces the number of accessed ways without hurting the performance. The PS-Cache takes advantage of the private-shared knowledge of the referenced block to reduce energy by accessing only those ways holding the kind of block looked up. Experimental results show that, on average, the PS-Cache architecture can reduce the dynamic energy consumption of L1 and L2 caches by 22 and 40%, respectively., This work has been jointly supported by the MINECO and European Commission (FEDER funds) under the project TIN2012-38341-C04-01 and the Fundaci’on Seneca-Agencia de Ciencia y Tecnología de la Región de Murcia under the project Jóvenes Líderes en Investigación 18956/JLI/13.
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Programmability and portability for exascale: Top down programming methodology and tools with StarSs

  • Subotic, Vladimir
  • Brinkmann, Steffen
  • Marjanovic, Vladimir
  • Badia, Rosa M.
  • Gracia, Jose
  • Niethammer, Christoph
  • Ayguade, Eduard
  • Labarta, Jesús
  • Valero, Mateo
StarSs is a task-based programming model that allows to parallelize sequential applications by means of annotating the code with compiler directives. The model further supports transparent execution of designated tasks on heterogeneous platforms, including clusters of GPUs. This paper focuses on the methodology and tools that complements the programming model forming a consistent development environment with the objective of simplifying the live of application developers.The programming environment includes the tools TAREADOR and TEMANEJO, which have been designed specifically for StarSs. TAREADOR, a Valgrind-based tool, allows a top-down development approach by assisting the programmer in identifying tasks and their data-dependencies across all concurrency levels of an application. TEMANEJO is a graphical debugger supporting the programmer by visualizing the task dependency tree on one hand, but also allowing to manipulate task scheduling or dependencies. These tools are complemented with a set of performance analysis tools (Scalasca, Cube and Paraver) that enable to fine tune StarSs application. © 2013 Elsevier B.V., We thankfully acknowledge the support of the European Commission through the TEXT project (FP7-261580) and the HiPEAC-3 Network of Excellence (FP7-ICT 287759), and the support of the Spanish Ministry of Education (TIN2007-60625, TIN2012-34557 and CSD2007-00050), the Generalitat de Catalunya (2009-SGR-980), Peer Reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Dense matrix computations on NUMA architectures with distance-aware work stealing

  • Al-Omairy, Rabab
  • Miranda, Guillermo
  • Ltaief, Hatem
  • Badia, Rosa M.
  • Martorell, Xavier
  • Labarta, Jesús
  • Keyes, David
We employ the dynamic runtime system OmpSs to decrease the overhead of data motion in the now ubiquitous non-uniform memory access (NUMA) high concurrency environment of multicore processors. The dense numerical linear algebra algorithms of Cholesky factorization and symmetric matrix inversion are employed as representative benchmarks. Work stealing occurs within an innovative NUMA-aware scheduling policy to reduce data movement between NUMA nodes. The overall approach achieves separation of concerns by abstracting the complexity of the hardware from the end users so that high productivity can be achieved. Performance results on a large NUMA system outperform the state-of-the-art existing implementations up to a twofold speedup for the Cholesky factorization, as well as the symmetric matrix inversion, while the OmpSs-enabled code maintains strong similarity to its original sequential version., The authors would like to thank the National Institute for Computational Sciences for granting us access on the Nautilus system. The KAUST authors acknowledge support of the Extreme Computing Research Center. The BSC-affiliated authors thankfully acknowledges the support of the European Commission through the HiPEAC-3 Network of Excellence (FP7-ICT 287759), Intel-BSC Exascale Lab and IBM/BSC Exascale Initiative collaboration, Spanish Ministry of Education (FPU), Computación de Altas Prestaciones VI (TIN2012-34557), Generalitat de Catalunya (2014-SGR-1051) and the grant SEV-2011-00067 of the Severo Ochoa Program.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

On the usefulness of object tracking techniques in performance analysis

  • Llort Sánchez, Germán
  • Servat, Harald
  • González García, Juan
  • Giménez Lucas, Judit
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
Understanding the behavior of a parallel application is crucial if we are to tune it to achieve its maximum performance. Yet the behavior the application exhibits may change over time and depend on the actual execution scenario: particular inputs and program settings, the number of processes used, or hardware-specific problems. So beyond the details of a single experiment a far more interesting question arises: how does the application behavior respond to changes in the execution conditions? In this paper, we demonstrate that object tracking concepts from computer vision have huge potential to be applied in the context of performance analysis. We leverage tracking techniques to analyze how the behavior of a parallel application evolves through multiple scenarios where the execution conditions change. This method provides comprehensible insights on the influence of different parameters on the application behavior, enabling us to identify the most relevant code regions and their performance trends. Copyright 2013 ACM., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Cross-stratum orchestration and flexgrid optical networks for datacenter federations

  • Velasco Esteban, Luis Domingo|||0000-0002-7345-296X
  • Asensio Garcia, Adrian|||0000-0002-3345-6093
  • Castro Casales, Alberto
  • Berral García, Josep Lluís|||0000-0003-3037-3580
  • Carrera Pérez, David|||0000-0003-4898-3424
  • López, Víctor
  • Fernández Palacios, Juan Pedro
Current inter-data-center connections are configured as static big fat pipes, which entails large bit rate over-provisioning and thus high operational costs for DC operators. On the other hand, network operators cannot share such connections between customers, because DC traffic varies greatly over time. Those connections are used to perform virtual machine migration and database synchronization among federated DCs, allowing elastic DC operations. To improve resource utilization and save costs, dynamic inter-DC connectivity is currently being targeted from a research point of view and in standardization form. In this article, we show that dynamic connectivity is not enough to guarantee elastic DC operations and might lead to poor performance provided that not enough overprovisioning of network resources is performed. To alleviate it to some extent, we propose using the flexgrid optical technology that enables finer spectrum granularity adaptation and the ability to dynamically increase and decrease the amount of optical resources assigned to connections. DCs can be interconnected through a flexgrid-based network controlled using a centralized software defined network, based on the architecture currently being proposed by the IETF; a cross-stratum orchestrator architecture coordinates DC and network elastically. Illustrative results show that dynamic elastic connectivity provides benefits by reducing the amount of overprovisioned network resources and facilitating elastic DC operations., The research leading to these results has received funding from the European Community’s Seventh Framework Pro- gramme FP7/2007-2013 under grant agreement no. 317999 IDEALIST project. Moreover, it was supported by the Span- ish science ministry through the TEC2011-27310 ELASTIC project.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

APPLE: Adaptive performance-predictable low-energy caches for reliable hybrid voltage operation

  • Maric, Bojan
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
  • Valero Cortés, Mateo|||0000-0003-2917-2482
Semiconductor technology evolution enables the design of resource-constrained battery-powered ultra-low-cost chips required for new market segments such as environment, urban life and body monitoring. Caches have been shown to be the main energy and area consumer in those chips. This paper proposes simple, hybrid-operation (high Vcc, ultra-low Vcc), single-Vcc domain Adaptive Performance- Predictable Low-Energy (APPLE) L1 cache designs based on replacing energy-hungry SRAM cells by more energy-efficient and smaller cells enhanced with extra cache lines set up in an adapted victim cache to still enable strong performance guarantees. APPLE caches are proven to largely outperform existing solutions in terms of energy and area efficiency.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Performance analysis and predictability of the software layer in Dynamic Binary Translators/Optimizers

  • Brankovic, Aleksandar
  • Stavrou, K.
  • Gibert Codina, Enric
  • Gonzalez, Antonio
Dynamic Binary Translators and Optimizers (DBTOs) have been established as a common architecture during the last years. They are used in many different systems, such as emulation, instrumentation tools and innovative HW/SW co-designed microarchitectures. Although many researchers worked on characterizing and reducing the emulation over-head, there are no published results that explain how the DBTO behaves from the microarchitectural prospective and how its behavior may be predicted based on high-level, guest application statistics. Such results are important for guiding design decisions and system optimization. In this paper we study the DBTO as an independent application by dividing its functionality into modules. We show that the behavior of the DBTO is not constant at all. The contribution of the different modules in the total overhead, the overhead itself, the microarchitectural interaction with the emulated application and the microarchitectural proffle of the different modules changes significantly based on the emulated application. This result comes in contrast to numerous papers that consider this behavior constant and exclude the DBTO from the simulation. Throughout this paper we detail this variance, we quantify it and we explain the reasons behind it. The insights presented in this work can be exploited towards the design of more efficient DBTOs and their early performance evaluation., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Tessellation: Refactoring the OS around explicit resource containers with continuous adaptation

  • Colmenares, Juan A.
  • Eads, Gage
  • Hofmeyry, Steven
  • Bird, Sarah L.
  • Moretó Planas, Miquel|||0000-0002-9848-8758
  • Chou, David
  • Gluzman, Brian
  • Roman, Eric
  • Bartolini, Davide B.
  • Mor, Nitesh
  • Asanovic, Krste
  • Kubiatowicz, John D.
Adaptive Resource-Centric Computing (ARCC) enables a simultaneous mix of high-throughput parallel, real-time, and interactive applications through automatic discovery of the correct mix of resource assignments necessary to achieve application requirements. This approach, embodied in the Tessellation manycore operating system, distributes resources to QoS domains called cells. Tessellation separates global decisions about the allocation of resources to cells from application-specific scheduling of resources within cells. We examine the implementation of ARCC in the Tessellation OS, highlight Tessellation's ability to provide predictable performance, and investigate the performance of Tessellation services within cells., This research is supported by Microsoft (Award #024263), Intel (Award #024894), matching U.C. Discovery funding (Award #DIG07-102270), and DOE ASCR FastOS Grant #DE-FG02-08ER25849. Additional support comes from Par Lab aliates National Instruments, Nokia, NVIDIA, Oracle, and Samsung. No part of this paper repre- sents views and opinions of the sponsors mentioned above. We thank other Par Lab members for their collaboration and feedback. J. A. Colmenares participated in this work while he was a post-doctoral scholar at UC Berkeley. M. Moreto is supported by a MEC/Fulbright Fellowship., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

DTM: degraded test mode for fault-aware probabilistic timing analysis

  • Slijepcevic, Mladen
  • Kosmidis, Leonidas
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
  • Quiñones, Eduardo
  • Cazorla Almeida, Francisco Javier
Existing timing analysis techniques to derive Worst-Case Execution Time (WCET) estimates assume that hardware in the target platform (e.g., the CPU) is fault-free. Given the performance requirements increase in current Critical Real-Time Embedded Systems (CRTES), the use of high-performance features and smaller transistors in current and future hardware becomes a must. The use of smaller transistors helps providing more performance while maintaining low energy budgets, however, hardware fault rates increase noticeably, affecting the temporal behaviour of the system in general, and WCET in particular. In this paper, we reconcile these two emergent needs of CRTES, namely, tight (and trustworthy) WCET estimates and the use of hardware implemented with smaller transistors. To that end we propose the Degraded Test Mode (DTM) that, in combination with fault-tolerant hardware designs and probabilistic timing analysis techniques, (i) enables the computation of tight and trustworthy WCET estimates in the presence of faults, (ii) provides graceful average and worst-case performance degradation due to faults, and (iii) requires modifications neither in WCET analysis tools nor in applications. Our results show that DTM allows accounting for the effect of faults at analysis time with low impact in WCET estimates and negligible hardware modifications., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

A proposal for task-generating loops in OpenMP

  • Teruel, Xavier
  • Klemm, Michael
  • Li, Kelvin
  • Martorell Bofill, Xavier|||0000-0002-0417-3430
  • Olivier, Stephen
  • Terboven, Christian
With the addition of the OpenMP* tasking model, programmers are able to improve and extend the parallelization opportunities of their codes. Programmers can also distribute the creation of tasks using a worksharing construct, which allows the generation of work to be parallelized. However, while it is possible to create tasks inside worksharing constructs, it is not possible to distribute work when not all threads reach the same worksharing construct. We propose a new worksharing-like construct that removes this restriction: the taskloop construct. With this new construct, we can distribute work when executing in the context of an explicit task, a single, or a master construct, enabling us to explore new parallelization opportunities in our applications. Although we focus our current work on evaluating expressiveness rather than performance evaluation, we present some initial performance results using a naive implementation for the new taskloop construct based on a lazy task instantiation mechanism., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Applying measurement-based probabilistic timing analysis to buffer resources

  • Kosmidis, Leonidas
  • Vardanega, Tulio
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
  • Quiñones, Eduardo
  • Cazorla Almeida, Francisco Javier
The use of complex hardware makes it difficult for current timing analysis techniques to compute trustworthy and tight worst-case execution time (WCET) bounds. Those techniques require detailed knowledge of the internal operation and state of the platform, at both the software and hardware level. Obtaining that information for modern hardware platforms is increasingly difficult. Measurement-Based Probabilistic Timing Analysis (MBPTA) reduces the cost of acquiring the knowledge needed for computing trustworthy and tight WCET bounds. MBPTA based on Extreme Value Theory requires the execution time of processor instructions to be independent and identically distributed (i.i.d.), which can be achieved with some hardware support. Previous proposals show how those properties can be achieved for caches. This paper considers, for the first time, the implications on MBPTA of using buffer resources. Buffers in general, and firstcome first-served (FCFS) buffers in particular, are of paramount importance as the complexity of hardware increases, since they allow managing contention in those resources where multiple requests may be pending. We show how buffers can be used in the context of MBPTA and provide illustrative examples., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Heterogeneous tasking on SMP/FPGA SoCs: The case of OmpSs and the Zynq

  • Filgueras Izquierdo, Antonio|||0000-0001-6071-3254
  • Gil Blasco, Eduard
  • Álvarez Martínez, Carlos|||0000-0003-0536-5183
  • Jiménez González, Daniel|||0000-0001-6064-7883
  • Martorell Bofill, Xavier|||0000-0002-0417-3430
  • Langer, Jan
  • Noguera Serra, Juan José
OmpSs is a directive-based programming model that uses OpenMP-like directives, that allow to execute the tasks annotated on both the SMPs and as FPGA kernels on modern SoC processors, like the Xilinx Zynq platform. OmpSs includes the support for accelerators (MIC, GPUs, FPGAs) and task dependencies, like OpenMP 4.0 will support. In this paper we present our approach for the support of FPGAs and the Zynq SoC, the current status of the implementation, its analysis and performance evaluation.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Multi-level unified caches for probabilistically time analysable real-time systems

  • Kosmidis, Leonidas
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
  • Quiñones, Eduardo
  • Cazorla Almeida, Francisco Javier
Caches are key resources in high-end processor architectures to increase performance. In fact, most high-performance processors come equipped with a multi-level cache hierarchy. In terms of guaranteed performance, however, cache hierarchies severely challenge the computation of tight worst-case execution time (WCET) estimates. On the one hand, the analysis of the timing behaviour of a single level of cache is already challenging, particularly for data accesses. On the other hand, unifying data and instructions in each level, makes the problem of cache analysis significantly more complex requiring tracking simultaneously data and instruction accesses to cache. In this paper we prove that multi-level cache hierarchies can be used in the context of Probabilistic Timing Analysis and tight WCET estimates can be obtained. Our detailed analysis (1) covers unified data and instruction caches, (2) covers different cache-write policies (write-through and write back), write allocation policies (write-allocate and non-write-allocate) and several inclusion mechanisms (inclusive, non-inclusive and exclusive caches), and (3) scales to an arbitrary number of cache levels. Our results show that the probabilistic WCET (pWCET) estimates provided by our analysis technique effectively benefit from having multi-level caches. For a two-level cache configuration and for EEMBC benchmarks, pWCET reductions are 55% on average (and up to 90%) with respect to a processor with a single level of cache., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Energy efficiency vs. performance of the numerical solution of PDEs: an application study on a low-power ARM-based cluster

  • Göddeke, Dominik
  • Komatitsch, D.
  • Geveler, Markus
  • Ribbrock, D.
  • Rajovic, Nikola
  • Puzovic, Nikola
  • Ramírez Bellido, Alejandro
Power consumption and energy efficiency are becoming critical aspects in the design and operation of large scale HPC facilities, and it is unanimously recognised that future exascale supercomputers will be strongly constrained by their power requirements. At current electricity costs, operating an HPC system over its lifetime can already be on par with the initial deployment cost. These power consumption constraints, and the benefits a more energy-efficient HPC platform may have on other societal areas, have motivated the HPC research community to investigate the use of energy-efficient technologies originally developed for the embedded and especially mobile markets. However, lower power does not always mean lower energy consumption, since execution time often also increases. In order to achieve competitive performance, applications then need to efficiently exploit a larger number of processors. In this article, we discuss how applications can efficiently exploit this new class of low-power architectures to achieve competitive performance. We evaluate if they can benefit from the increased energy efficiency that the architecture is supposed to achieve. The applications that we consider cover three different classes of numerical solution methods for partial differential equations, namely a low-order finite element multigrid solver for huge sparse linear systems of equations, a Lattice-Boltzmann code for fluid simulation, and a high-order spectral element method for acoustic or seismic wave propagation modelling. We evaluate weak and strong scalability on a cluster of 96 ARM Cortex-A9 dual-core processors and demonstrate that the ARM-based cluster can be more efficient in terms of energy to solution when executing the three applications compared to an x86-based reference machine., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

EcoTM: Conflict-aware economical unbounded hardware transactional memory

  • Tomić, Saša
  • Akpinar, Ege
  • Cristal Kestelman, Adrián|||0000-0003-1277-9296
  • Unsal, Osman Sabri
  • Valero Cortés, Mateo|||0000-0003-2917-2482
Transactional Memory (TM) is a promising paradigm for parallel programming. TM allows a thread to make a series of memory accesses as a single, atomic, transaction, while avoiding deadlocks, livelocks, and other problems commonly associated with lock-based programming. In this paper we explore Hardware support for TM (HTM). In particular, we explore how HTM can efficiently support transactions of nearly unlimited size. For this purpose we propose EcoTM, an economical unbounded HTM that improves the efficiency of conflict detection between very large transactions by activating conflict-detection logic only for potentially-conflicting locations: shared and speculatively modified. EcoTM detects the potentially-conflicting locations automatically, without any program annotations. We evaluate EcoTM performance by comparing it with ideal-lazy HTM, unbounded eager HTM with perfect signatures, and LogTM-SE. Our evaluations show that EcoTM has similar performance as the ideal-lazy HTM, 8.8% better than the eager- perfect HTM, and over 35.7% better than LogTM-SE, on the average., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Identifying critical code sections in dataflow programming models

  • Subotic, Vladimir
  • Sancho, Jose Carlos
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
  • Valero Cortés, Mateo|||0000-0003-2917-2482
The years of practice in optimizing applications point that the major issue is focus - identifying the critical code section whose optimization would yield the highest overall speedup. While this issue is mainly solved for sequential applications, it remains a serious hurdle in the world of parallel computing. Furthermore, the newest dataflow parallel programming models expose very irregular parallelism, making the identification of the critical code section even harder. To address this issue, we designed an environment that identifies critical code sections in applications. The programmer can use this environment to estimate the potential benefits of the optimization for a specific parallel platform. This is very important because the programmer can anticipate the benefits of his optimization and assure that the optimization is worth the effort. Furthermore, we showed that in many applications, the choice of the critical code section decisively depends on the configuration of the target machine. For instance, in HP Linpack, optimizing a task that takes 0.49% of the total computation time yields the overall speedup of less than 0.25% on one machine, and at the same time, yields the overall speedup of more than 24% on a machine with different number of cores.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Parallel algorithm for simulating the spatial transmission of influenza in EpiGraph

  • Martín, Gonzalo
  • Singh, David E.
  • Marinescu, Maria-Cristina
  • Carretero, Jesús
This paper introduces an approach to modeling and simulating the propagation of u-like infectious diseases over large, widely spread urban areas connected by transportation networks. We incorporate geographic location and a transportation model into our region-based, closed-world EpiGraph simulator to realistically model the movement of the virus between different geographic regions. The resulting simulator can assist in understanding how outbreaks propagate between far apart regions due to the movement of people outside their base location. This paper describes the MPI-based implementation of EpiGraph and its performance evaluation when simulating large-scale scenarios. We evaluate the simulator both on a distributed memory system and on a shared memory system., This work has been partially supported by the Spanish Ministry of Science under the grant IPT-430000-2010-14., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

OFAR-CM: Efficient Dragonfly networks with simple congestion management

  • García, Marina
  • Vallejo, Enrique
  • Beivide Palacio, Ramon
  • Valero Cortés, Mateo|||0000-0003-2917-2482
  • Rodríguez, Germán
Dragonfly networks are appealing topologies for large-scale Data center and HPC networks, that provide high throughput with low diameter and moderate cost. However, they are prone to congestion under certain frequent traffic patterns that saturate specific network links. Adaptive non-minimal routing can be used to avoid such congestion. That kind of routing employs longer paths to circumvent local or global congested links. However, if a distance-based deadlock avoidance mechanism is employed, more Virtual Channels (VCs) are required, what increases design complexity and cost. OFAR (On-the-Fly Adaptive Routing) is a previously proposed routing that decouples VCs from deadlock avoidance, making local and global misrouting affordable. However, the severity of congestion with OFAR is higher, as it relies on an escape sub network with low bisection bandwidth. Additionally, OFAR allows for unlimited misroutings on the escape sub network, leading to unbounded paths in the network and long latencies. In this paper we propose and evaluate OFAR-CM, a variant of OFAR combined with a simple congestion management (CM) mechanism which only relies on local information, specifically the credit count of the output ports in the local router. With simple escape sub networks such as a Hamiltonian ring or a tree, OFAR outperforms former proposals with distance-based deadlock avoidance. Additionally, although long paths are allowed in theory, in practice packets arrive at their destination in a small number of hops. Altogether, OFAR-CM constitutes the first practicable mechanism to the date that supports both local and global misrouting in Dragonfly networks., The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. ERC-2012-Adg-321253- RoMoL, the Spanish Ministry of Science under contracts TIN2010-21291-C02-02, TIN2012-34557, and by the European HiPEAC Network of Excellence. M. García participated in this work while affiliated with the University of Cantabria., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Improving communication in PGAS environments: Static and dynamic coalescing in UPC

  • Alvanos, Michail
  • Farreras Esclusa, Montserrat|||0000-0003-2027-1919
  • Tiotto, Ettore
  • Amaral, José Nelson
  • Martorell Bofill, Xavier|||0000-0002-0417-3430
The goal of Partitioned Global Address Space (PGAS) languages is to improve programmer productivity in large scale parallel machines. However, PGAS programs may have many fine-grained shared accesses that lead to performance degradation. Manual code transformations or compiler optimizations are required to improve the performance of programs with fine-grained accesses. The downside of manual code transformations is the increased program complexity that hinders programmer productivity. On the other hand, most compiler optimizations of fine-grain accesses require knowledge of physical data mapping and the use of parallel loop constructs. This paper presents an optimization for the Unified Parallel C language that combines compile time (static) and runtime (dynamic) coalescing of shared data, without the knowledge of physical data mapping. Larger messages increase the network efficiency and static coalescing decreases the overhead of library calls. The performance evaluation uses two microbenchmarks and three benchmarks to obtain scaling and absolute performance numbers on up to 32768 cores of a Power 775 machine. Our results show that the compiler transformation results in speedups from 1.15X up to 21X compared with the baseline versions and that they achieve up to 63% the performance of the MPI versions., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

On the selection of adder unit in energy efficient vector processing

  • Ratkovic, Ivan
  • Palomar Pérez, Óscar|||0000-0001-6729-4187
  • Stanic, Milan
  • Unsal, Osman Sabri
  • Cristal Kestelman, Adrián|||0000-0003-1277-9296
  • Valero Cortés, Mateo|||0000-0003-2917-2482
Vector processors are a very promising solution for mobile devices and servers due to their inherently energy-efficient way of exploiting data-level parallelism. Previous research on vector architectures predominantly focused on performance, so vector processors require a new design space exploration to achieve low power. In this paper, we present a design space exploration of adder unit for vector processors (VA), as it is one of the crucial components in the core design with a non-negligible impact in overall performance and power. For this interrelated circuit-architecture exploration, we developed a novel framework with both architectural- and circuit-level tools. Our framework includes both design- (e.g. adder's family type) and vector architecture-related parameters (e.g. vector length). Finally, we present guidelines on the selection of the most appropriate VA for different types of vector processors according to different sets of metrics of interest. For example, we found that 2-lane configurations are more EDP (Energy×Delay)-efficient than single lane configurations for low-end mobile processors., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Capítulo o parte de un libro (BookPart)

FLEX-MPI: An MPI Extension for Supporting Dynamic Load Balancing on Heterogeneous Non-dedicated Systems

  • Martin, Gonzalo
  • Marinescu, Maria-Cristina
  • Singh, David E.
  • Carretero, Jesús
This paper introduces FLEX-MPI, a novel runtime approach for the dynamic load balancing of MPI-based SPMD applications running on heterogeneous platforms in the presence of dynamic external loads. To effectively balance the workload, FLEX-MPI monitors the actual performance of applications via hardware counters and the MPI profiling interface—with a negligible overhead and minimal code modifications. Our results show that by using this approach the execution time of an application may be significantly reduced., This work has been partially supported by the Spanish Ministry of Science under the grant IPT-430000-2010-14., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Parallelizing general histogram application for CUDA architectures

  • Milic, Ugljesa
  • Gelado Fernandez, Isaac
  • Puzovic, Nikola
  • Ramírez Bellido, Alejandro
  • Tomasevic, Milo
Histogramming is a tool commonly used in data analysis. Although its serial version is simple to implement, providing an efficient and scalable way to parallelize it can be challenging. This especially holds in case of platforms that contain one or several massively parallel devices like CUDA-capable GPUs due to issues with domain decomposition, use of global memory and similar. In this paper we compare two approaches for implementing general purpose histogramming on GPUs. The first algorithm is based on private copies of bin counters stored in shared memory for each block of threads. The second one uses the Thrust library to sort the input elements and then to search for upper bounds according to bin widths. For both algorithms we analyze how the speedup over the sequential version depends on the size of input collection, number of bins, and the type and distribution of input elements. We also implement overlapping of data transfers between host CPU and CUDA device with kernel execution. For both algorithms we analyze the pros and cons in detail. For example, privatization strategy can be up to 2x faster than sort-search with realistic inputs, but can only support a limited number of bins. On the other hand, sort-search strategy has about 50% higher speedup than privatization when we use characters as input and can support unlimited number of bins. Finally, we perform an exploration to determine the optimal algorithm depending on the characteristics and values of input parameters., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Improving the energy efficiency of hardware-assisted watchpoint systems

  • Karakostas, Vasileios
  • Tomić, Saša
  • Unsal, Osman Sabri
  • Nemirovsky, Mario
  • Cristal Kestelman, Adrián|||0000-0003-1277-9296
Hardware-assisted watchpoint systems enhance the execution of numerous dynamic software techniques, such as memory protection, module isolation, deterministic execution, and data race detection. In this paper, we show that previous hardware proposals may introduce significant energy overheads, and propose WatchPoint Filtering (WPF), a novel filtering mechanism that eliminates unnecessary watchpoint checks. We evaluate WPF on two state-of-the-art proposals for hardware-assisted watchpoints using two common memory checkers. WPF eliminates 83% of the watchpoint checks (up to 99.7%) and reduces 57% of the dynamic energy overhead (up to 78%) on average, without introducing additional performance execution overhead.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Supercomputing with commodity CPUs: are mobile SoCs ready for HPC?

  • Rajovic, Nikola
  • Carpenter, Paul Matthew
  • Gelado Fernandez, Isaac
  • Puzovic, Nikola
  • Ramírez Bellido, Alejandro
  • Valero Cortés, Mateo|||0000-0003-2917-2482
Best Student Paper Award, a la SC13: International Conference for High Performance Computing, Networking, Storage and Analysis, In the late 1990s, powerful economic forces led to the adoption of commodity desktop processors in high-performance computing. This transformation has been so effective that the June 2013 TOP500 list is still dominated by x86. In 2013, the largest commodity market in computing is not PCs or servers, but mobile computing, comprising smart-phones and tablets, most of which are built with ARM-based SoCs. This leads to the suggestion that once mobile SoCs deliver sufficient performance, mobile SoCs can help reduce the cost of HPC. This paper addresses this question in detail. We analyze the trend in mobile SoC performance, comparing it with the similar trend in the 1990s. We also present our experience evaluating performance and efficiency of mobile SoCs, deploying a cluster and evaluating the network and scalability of production applications. In summary, we give a first answer as to whether mobile SoCs are ready for HPC., Peer Reviewed, Award-winning
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Towards general purpose computations on low-end mobile GPUs

  • Trompouki, Matina M.
  • Kosmidis, Leonidas
GPUs traditionally offer high computational capabilities, frequently higher than their CPU counterparts. While high-end mobile GPUs vendors introduced recently general purpose APIs, such as OpenCL, to leverage their computational power, the vast majority of the mobile devices lack such support. Despite that their graphics APIs have similarities with desktop graphics APIs, they have significant differences, which prevent the use of well-known techniques that offer general-purpose computations over such interfaces. In this paper we show how these obstacles can be overcome, in order to achieve general purpose programmability of these devices. As a proof of concept we implemented our proposal on a real embedded platform (Raspberry Pi) based on Broadcom's VideoCore IV GPU, obtaining a speedup of 7.2× over the CPU., This work has been partially supported by the Spanish Ministry of Science and Innovation under grant TIN2015-65316-P and the HiPEAC Network of Excellence. Leonidas Kosmidis is also funded by the Spanish Ministry of Education under the FPU grant AP2010-4208.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Enabling distributed key-value stores with low latency-impact snapshot support

  • Polo Bardés, Jordà
  • Becerra Fontal, Yolanda|||0000-0003-2357-7796
  • Carrera Pérez, David|||0000-0003-4898-3424
  • Torres Viñals, Jordi|||0000-0003-1963-7418
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
  • Spreitzer, Mike
  • Steinder, Malgorzata
Current distributed key-value stores generally provide greater scalability at the expense of weaker consistency and isolation. However, additional isolation support is becoming increasingly important in the environments in which these stores are deployed, where different kinds of applications with different needs are executed, from transactional workloads to data analytics. While fully-fledged ACID support may not be feasible, it is still possible to take advantage of the design of these data stores, which often include the notion of multiversion concurrency control, to enable them with additional features at a much lower performance cost and maintaining its scalability and availability. In this paper we explore the effects that additional consistency guarantees and isolation capabilities may have on a state of the art key-value store: Apache Cassandra. We propose and implement a new multiversioned isolation level that provides stronger guarantees without compromising Cassandra's scalability and availability. As shown in our experiments, our version of Cassandra allows Snapshot Isolation-like transactions, preserving the overall performance and scalability of the system., This work is partially supported by the Ministry of Science and Technology of Spain and the European Union’s FEDER funds (TIN2007-60625, TIN2012-34557), by the Generalitat de Catalunya (2009-SGR-980), by the BSC-CNS Severo Ochoa program (SEV-2011-00067), by the HiPEAC European Network of Excellence (IST- 004408, FP7-ICT-217068, FP7-ICT-287759), and by IBM through the 2008 and 2010 IBM Faculty Award program., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Aeneas: A tool to enable applications to effectively use non-relational databases

  • Cugnasco, Cesare
  • Hernández, Roger
  • Becerra Fontal, Yolanda|||0000-0003-2357-7796
  • Torres Viñals, Jordi|||0000-0003-1963-7418
  • Ayguadé Parra, Eduard|||0000-0002-5146-103X
Non-relational databases arise as a solution to solve the scalability problems of relational databases when dealing with big data applications. However, they are highly configurable prone to user decisions that can heavily affect their performance. In order to maximize the performance, different data models and queries should be analyzed to choose the best fit. This may involve a wide range of tests and may result in productivity issues. We present Aeneas, a tool to support the design of data management code for applications using non-relational databases. Aeneas provides an easy and fast methodology to support the decision about how to organize and retrieve data in order to improve the performance., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Circuit design of a novel adaptable and reliable L1 data cache

  • Seyedi, Azam
  • Yalcin, Gulay
  • Unsal, Osman Sabri
  • Cristal Kestelman, Adrián|||0000-0003-1277-9296
This paper proposes a novel adaptable and reliable L1 data cache design (Adapcache) with the unique capability of automatically adapting itself for different supply voltage levels and providing the highest reliability. Depending on the supply voltage level, Adapcache defines three operating modes: In high supply voltages, Adapcache provides reliability through single-bit parity. In middle range of supply voltages, Adapcache writes data to two separate cache-lines simultaneously in order to use one line for error recovery when the other line is faulty. In near threshold supply voltages, Adapcache writes data to three separate cache-lines simultaneously in order to provide the correct data based on bitwise majority voter. We design and simulate one embodiment of the Adapcache as a 64-KB L1 data cache with 45-nm CMOS technology at 2GHz processor frequency for almost nominal supply voltages (1V-0.6V), at 900MHz for middle supply voltages (0.6V-0.4V), and at 400MHz for near threshold supply voltages (0.4V-0.32V). According to our experimental results, the energy reduction and latency as well as cache capacity usage are improved compared to typical previous proposals, Triple Modular Redundancy (TMR) and Double Modular Redundancy (DMR) techniques and also to the state of the art proposal, Parichute Error Correction Code (ECC).
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Improving performance of all-to-all communication through loop scheduling in PGAS environments

UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

An OpenMP* barrier using SIMD instructions for Intel® Xeon Phi™ coprocessor

  • Caballero, Diego
  • Duran González, Alejandro
  • Martorell Bofill, Xavier|||0000-0002-0417-3430
Barrier synchronisation is a widely-studied topic since the supercomputer era due to its significant impact on the overall performance of parallel applications. With the current shift to many-core architectures, such as the Intel® Many Integrated Core Architecture, software barriers need to be revisited from an on-chip point of view to exploit their new specific resources. In this paper, we propose a tree-based barrier that takes advantage of SIMD instructions and the inter-thread cache locality provided by the 4-way SMT of the Intel® Xeon PhiTM coprocessor. Our SIMD approach shows a speed-up of up to 2.84x over the default Intel OpenMP* barrier in the EPCC barrier microbenchmark. It also improves by up to 60% and 21% the Livermore Loop kernel number six and the NAS MG benchmark, respectively., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Framework for a productive performance optimization

  • Servat, Harald
  • Llort, German
  • Huck, Kevin A.
  • Giménez Lucas, Judit
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
Modern supercomputers deliver large computational power, but it is difficult for an application to exploit such power. One factor that limits the application performance is the single node performance. While many performance tools use the microprocessor performance counters to provide insights on serial node performance issues, the complex semantics of these counters pose an obstacle to an inexperienced developer. We present a framework that allows easy identification and qualification of serial node performance bottlenecks in parallel applications. The output of the framework is precise and it is capable of correlating performance inefficiencies with small regions of code within the application. The framework not only points to regions of code but also simplifies the semantics of the performance counters into metrics that refer to processor functional units. With such information the developer can focus on the identified code and improve it by knowing which processor execution unit is degrading the performance. To demonstrate the usefulness of the framework we apply it to three already optimized applications using realistic inputs and, according to the results, modify their source code. By doing modifications that require little effort, we successfully increase the applications’ performance from 10% to 30% and thus shorten the time required to reach the solution and/or allow facing increased problem sizes., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Graphene-enabled wireless communication for massive multicore architectures

  • Abadal Cavallé, Sergi|||0000-0003-0941-0260
  • Alarcón Cot, Eduardo José|||0000-0001-7663-7153
  • Cabellos Aparicio, Alberto|||0000-0001-9329-7584
  • Lemme, Max
  • Nemirovsky, Mario
Doctoral Program Student Honor Programme, de l'empresa ENTEL, Current trends in microprocessor architecture design are leading towards a dramatic increase of core-level parallelization, wherein a given number of independent processors or cores are interconnected. Since the main bottleneck is foreseen to migrate from computation to communication, efficient and scalable means of inter-core communication are crucial for guaranteeing steady performance improvements in many-core processors. As the number of cores grows, it remains unclear whether initial proposals, such as the Network-on-Chip (NoC) paradigm, will meet the stringent requirements of this scenario. This position paper presents a new research area where massive multicore architectures have wireless communication capabilities at the core level. This goal is feasible by using graphene-based planar antennas, which can radiate signals at the Terahertz band while utilizing lower chip area than its metallic counterparts. To the best of our knowledge, this is the first work that discusses the utilization of graphene-enabled wireless communication for massive multicore processors. Such wireless systems enable broadcasting, multicasting, all-to-all communication, as well as significantly reduce many of the issues present in massively multicore environments, such as data coherency, consistency, synchronization and communication problems. Several open research challenges are pointed out related to implementation, communications and multicore architectures, which pave the way for future research in this multidisciplinary area., Peer Reviewed, Award-winning
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Producer-consumer: the programming model for future many-core processors

  • Prat Pérez, Arnau
  • Domínguez Sal, David
  • Larriba Pey, Josep|||0000-0002-7070-9256
  • Troncoso, Pedro
The massive addition of cores on a chip is adding more pressure to the accesses to main memory. In order to avoid this bottleneck, we propose the use of a simple producer-consumer model, which allows for the temporary results to be transferred directly from one task to another. These data transfer operations are performed within the chip, using on-chip memory, thus avoiding costly off-chip memory accesses. We implement this model on a real many-core processor, the 48-core Intel Single-chip Cloud Computer processor using its on-chip memory facilities. We find that the Producer-Consumer model adapts to such architectures and allow to achieve good task and data parallelism. For the evaluation of the proposed platform we implement a graph-based application using the Producer- Consumer model. Our tests show that the model scales very well as it takes advantage of the on-chip memory. The execution times of our implementation are up to 9 times faster than the baseline implementation, which relies on storing the temporary results to main memory., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Thread assignment of multithreaded network applications in multicore/multithreaded processors

  • Radojković, Petar
  • Cakarevic, Vladimir
  • Verdú Mulà, Javier|||0000-0003-4485-2419
  • Pajuelo González, Manuel Alejandro|||0000-0002-5510-6860
  • Cazorla Almeida, Francisco Javier
  • Nemirovsky, Mario
  • Valero Cortés, Mateo|||0000-0003-2917-2482
© 2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., The introduction of multithreaded processors comprised of a large number of cores with many shared resources makes thread scheduling, and in particular optimal assignment of running threads to processor hardware contexts to become one of the most promising ways to improve the system performance. However, finding optimal thread assignments for workloads running in state-of-the-art multicore/multithreaded processors is an NP-complete problem. In this paper, we propose BlackBox scheduler, a systematic method for thread assignment of multithreaded network applications running on multicore/multithreaded processors. The method requires minimum information about the target processor architecture and no data about the hardware requirements of the applications under study. The proposed method is evaluated with an industrial case study for a set of multithreaded network applications running on the UltraSPARC T2 processor. In most of the experiments, the proposed thread assignment method detected the best actual thread assignment in the evaluation sample. The method improved the system performance from 5 to 48 percent with respect to load balancing algorithms used in state-of-the-art OSs, and up to 60 percent with respect to a naive thread assignment., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Efficient routing mechanisms for Dragonfly networks

  • García, Marina
  • Vallejo, Enrique
  • Beivide Palacio, Julio Ramón
  • Odriozola, Miguel
  • Valero Cortés, Mateo|||0000-0003-2917-2482
High-radix hierarchical networks are cost-effective topologies for large scale computers. In such networks, routers are organized in super nodes, with local and global interconnections. These networks, known as Dragonflies, outperform traditional topologies such as multi-trees or tori, in cost and scalability. However, depending on the traffic pattern, network congestion can lead to degraded performance. Misrouting (non-minimal routing) can be employed to avoid saturated global or local links. Nevertheless, with the current deadlock avoidance mechanisms used for these networks, supporting misrouting implies routers with a larger number of virtual channels. This exacerbates the buffer memory requirements that constitute one of the main constraints in high-radix switches. In this paper we introduce two novel deadlock-free routing mechanisms for Dragonfly networks that support on-the-fly adaptive routing. Using these schemes both global and local misrouting are allowed employing the same number of virtual channels as in previous proposals. Opportunistic Local Misrouting obtains the best performance by providing the highest routing freedom, and relying on a deadlock-free escape path to the destination for every packet. However, it requires Virtual Cut-Through flow-control. By contrast, Restricted Local Misrouting prevents the appearance of cycles thanks to a restriction of the possible routes within super nodes. This makes this mechanism suitable for both Virtual Cut-Through and Wormhole networks. Evaluations show that the proposed deadlock-free routing mechanisms prevent the most frequent pathological issues of Dragonfly networks. As a result, they provide higher performance than previous schemes, while requiring the same area devoted to router buffers., This work has been supported by the Spanish Ministry of Science under contracts TIN2010-21291-C02-02, TIN2012-34557, and by the European HiPEAC Network of Excellence. The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. ERC-2012-Adg-321253-RoMoL. M. Garc´ıa and M. Odriozola participated in this research work while they were affiliated with the University of Cantabria., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Characterizing fault propagation in safety-critical processor designs

  • Espinosa, Jaime
  • Hernandez, Carles
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
Achieving reduced time-to-market in modern electronic designs targeting safety critical applications is becoming very challenging, as these designs need to go through a certification step that introduces a non-negligible overhead in the verification and validation process. To cope with this challenge, safety-critical systems industry is demanding new tools and methodologies allowing quick and cost-effective means for robustness verification. Microarchitectural simulators have been widely used to test reliability properties in different domains but their use in the process of robustness verification remains yet to be validated against other accepted methods such as RTL or gate-level simulation. In this paper we perform fault injections in an RTL model of a processor to characterize fault propagation. The results and conclusions of this characterization will serve to devise to what extent fault injection methodologies for robustness verification using microarchitectural simulators can be employed., The research leading to these results has received funding from the ARTEMIS Joint Undertaking VeTeSS project under grant agreement number 295311. This work has also been funded by the Ministry of Economy and Competitiveness of Spain under contract TIN2012-34557 and HiPEAC. Jaume Abella is partially supported by the Ministry of Economy and Competitiveness under Ramon y Cajal postdoctoral fellowship number RYC-2013-14717., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Business process mining from e-commerce web logs

  • Poggi Mastrokalo, Nicolas
  • Muthusamy, Vinod
  • Carrera Pérez, David|||0000-0003-4898-3424
  • Khalaf, Rania
The dynamic nature of the Web and its increasing importance as an economic platform create the need of new methods and tools for business efficiency. Current Web analytic tools do not provide the necessary abstracted view of the underlying customer processes and critical paths of site visitor behavior. Such information can offer insights for businesses to react effectively and efficiently. We propose applying Business Process Management (BPM) methodologies to e-commerce Website logs, and present the challenges, results and potential benefits of such an approach. We use the Business Process Insight (BPI) platform, a collaborative process intelligence toolset that implements the discovery of loosely-coupled processes, and includes novel process mining techniques suitable for the Web. Experiments are performed on custom click-stream logs from a large online travel and booking agency. We first compare Web clicks and BPM events, and then present a methodology to classify and transform URLs into events. We evaluate traditional and custom process mining algorithms to extract business models from real-life Web data. The resulting models present an abstracted view of the relation between pages, exit points, and critical paths taken by customers. Such models show important improvements and aid high-level decision making and optimization of e-commerce sites compared to current state-of-art Web analytics., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

HPC system software for regular and irregular parallel applications

  • Morari, Alessandro
  • Valero Cortés, Mateo|||0000-0003-2917-2482
The upcoming generation of system software for High Performance Computing is expected to provide a richer set of functionalities without compromising application performance. This Ph.D. thesis addresses the problem of designing scalable system software for both regular and irregular applications. The contributions are two-fold. First, we evaluate the drawbacks of current HPC system software for regular applications. We describe a methodology to precisely measure jitter on a general-purpose OS. Considering a lightweight operating system (IBM CNK), we analyze the overhead of adding support for a missing feature such as dynamic memory management. Second, we focus on irregular applications and build a specialized runtime system to enhance this kind of applications on common HPC flop intensive systems. The proposed runtime system provides a global address space abstraction of a distributed memory machine combined with a transparent fork/join execution model and it also includes lightweight multithreading and network message aggregation., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

PerfBound: Conserving Energy with Bounded Overheads in On/Off-Based HPC Interconnects

  • Saravanan, Karthikeyan P.
  • Carpenter, Paul Matthew
Energy and power are key challenges in high-performance computing. System energy efficiency must be significantly improved, and this requires greater efficiency in all subcomponents. An important target of optimization is the interconnect, since network links are always on, consuming power even during idle periods. A large number of HPC machines have a primary interconnect based on Ethernet (about 40 percent of TOP500 machines), which, since 2010, has included support for saving power via Energy Efficient Ethernet (EEE). Nevertheless, it is unlikely that HPC interconnects would use these energy saving modes unless the performance overhead is known and small. This paper presents PerfBound, a self-contained technique to manage on/off-based networks such as EEE, minimizing interconnect link energy consumption subject to a bound on the performance degradation. PerfBound does not require changes to the applications and it uses only local information already available at switches and NICs without introducing additional communication messages, and is also compatible with multi-hop networks. PerfBound is evaluated using traces from a production supercomputer. For twelve out of fourteen applications, PerfBound has high energy savings, up to 70 percent for only 1 percent performance degradation. This paper also presents DynamicFastwake, which extends PerfBound to exploit multiple low-power states. DynamicFastwake achieves an energy-delay product 10 percent lower than the original PerfBound technique, This research was supported by European Union’s 7th Framework Programme [FP7/2007-2013] under the Mont-Blanc-3 (FP7-ICT-671697) and EUROSERVER (FP7-ICT-610456) projects, the Ministry of Economy and Competitiveness of Spain (TIN2012-34557 and TIN2015-65316), Generalitat de Catalunya (FI-AGAUR 2012 FI B 00644, 2014-SGR-1051 and 2014-SGR-1272), the European Union’s Horizon2020 research and innovation programme under the HiPEAC-3 Network of Excellence (ICT-287759), and the Severo Ochoa Program (SEV-2011-00067) of the Spanish Government., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Data placement in HPC architectures with heterogeneous off-chip memory

  • Pavlovic, Milan
  • Puzovic, Nikola
  • Ramírez Bellido, Alejandro
The performance of HPC applications is often bounded by the underlying memory system's performance. The trend of increasing the number of cores on a chip imposes even higher memory bandwidth and capacity requirements. The limitations of traditional memory technologies are pushing research in the direction of hybrid memory systems that, besides DRAM, include one or more modules based on some of the higher-density non-volatile memory technologies, where one of them will provide the required bandwidth, while the other will provide the required capacity for the application. This creates many challenges with data placement and migration policies between the modules of such hybrid memory system. In this paper, we propose an architecture with a hybrid memory design that places two technologically different memory modules in a flat address space. On such system, we evaluate several HPC workloads against different data placement and migration policies, compare their performance by means of execution time and the number of non-volatile memory writes, and consider how it can be applied to the future HPC architectures. Our results show that the hybrid memory system with dynamic page migration and limited DRAM capacity, can achieve performance that is comparable to a hypothetical, hard to implement, DRAM-only system.
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

A hardware evaluation of cache partitioning to improve utilization and energy-efficiency while preserving responsiveness

  • Cook, Henry
  • Moretó Planas, Miquel|||0000-0002-9848-8758
  • Bird, Sarah L.
  • Dao, Khanh
  • Patterson, David
  • Asanovic, Krste
Computing workloads often contain a mix of interactive, latency-sensitive foreground applications and recurring background computations. To guarantee responsiveness, interactive and batch applications are often run on disjoint sets of resources, but this incurs additional energy, power, and capital costs. In this paper, we evaluate the potential of hardware cache partitioning mechanisms and policies to improve efficiency by allowing background applications to run simultaneously with interactive foreground applications, while avoiding degradation in interactive responsiveness. We evaluate these tradeoffs using commercial x86 multicore hardware that supports cache partitioning, and find that real hardware measurements with full applications provide different observations than past simulation-based evaluations. Co-scheduling applications without LLC partitioning leads to a 10% energy improvement and average throughput improvement of 54% compared to running tasks separately, but can result in foreground performance degradation of up to 34% with an average of 6%. With optimal static LLC partitioning, the average energy improvement increases to 12% and the average throughput improvement to 60%, while the worst case slowdown is reduced noticeably to 7% with an average slowdown of only 2%. We also evaluate a practical low-overhead dynamic algorithm to control partition sizes, and are able to realize the potential performance guarantees of the optimal static approach, while increasing background throughput by an additional 19%., We would especially like to thank everyone at Intel who made it possible for us to use the cache-partitioning machine in this paper, including Opher Kahn, Andrew Herdrich, Ravi Iyer, Gans Srinivasa, Mark Rowland, Ian Steiner and Henry Gabb. We would also like to Scott Beamer, Chris Celio, Shoaib Kamil, Leo Meyerovich, and David Sheeld for allowing us to study their applications. Additionally, we would like to thank our colleagues in the Par Lab for their continual advice, support, and, feedback. Research supported by Microsoft (Award 024263) and Intel (Award 024894) funding and by matching funding by U.C. Discovery (Award DIG07-10227). Additional support comes from ParLab aliates Nokia, NVIDIA, Oracle, and Samsung. M. Moreto was supported by the Spanish Ministry of Science under contract TIN2012-34557, a MEC/Fulbright Fellowship, and by an AGAUR award (BE-DGR 2010)
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Analysis and RTL correlation of instruction set simulators for automotive microcontroller robustness verification

  • Espinosa, Jaime
  • Hernandez, Carles
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
  • de Andres, David
  • Ruiz, Juan C.
Increasingly complex microcontroller designs for safety-relevant automotive systems require the adoption of new methods and tools to enable a cost-effective verification of their robustness. In particular, costs associated to the certification against the ISO26262 safety standard must be kept low for economical reasons. In this context, simulation-based verification using instruction set simulators (ISS) arises as a promising approach to partially cope with the increasing cost of the verification process as it allows taking design decisions in early design stages when modifications can be performed quickly and with low cost. However, it remains to be proven that verification in those stages provides accurate enough information to be used in the context of automotive microcontrollers. In this paper we analyze the existing correlation between fault injection experiments in an RTL microcontroller description and the information available at the ISS to enable accurate ISS-based fault injection., The research leading to these results has received funding from the ARTEMIS Joint Undertaking VeTeSS project under grant agreement number 295311. This work has also been funded by the Ministry of Science and Technology of Spain under contract TIN2012-34557 and HiPEAC. Jaume Abella is partially supported by the Ministry of Economy and Competitiveness under Ramon y Cajal postdoctoral fellowship number RYC-2013-14717., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Dense matrix computations on NUMA architectures with distance-aware work stealing

  • Al-Omairy, Rabab
  • Miranda Álamo, Guillermo
  • Ltaief, Hatem
  • Badia Sala, Rosa Maria|||0000-0003-2941-5499
  • Martorell Bofill, Xavier|||0000-0002-0417-3430
  • Labarta Mancho, Jesús José|||0000-0002-7489-4727
  • Keyes, David E.
We employ the dynamic runtime system OmpSs to decrease the overhead of data motion in the now ubiquitous non-uniform memory access (NUMA) high concurrency environment of multicore processors. The dense numerical linear algebra algorithms of Cholesky factorization and symmetric matrix inversion are employed as representative benchmarks. Work stealing occurs within an innovative NUMA-aware scheduling policy to reduce data movement between NUMA nodes. The overall approach achieves separation of concerns by abstracting the complexity of the hardware from the end users so that high productivity can be achieved. Performance results on a large NUMA system outperform the state-of-the-art existing implementations up to a two fold speedup for the Cholesky factorization, as well as the symmetric matrix inversion, while the OmpSs-enabled code maintains strong similarity to its original sequential version., The authors would like to thank the National Institute for Computational Sciences for granting us access on the Nautilus system. The KAUST authors acknowledge support of the Extreme Computing Research Center. The BSC-affiliated authors thankfully acknowledges the support of the European Commission through the HiPEAC-3 Network of Excellence (FP7-ICT 287759), Intel-BSC Exascale Lab and IBM/BSC Exascale Initiative collaboration, Spanish Ministry of Education (FPU), Computación de Altas Prestaciones VI (TIN2012-34557), Generalitat de Catalunya (2014-SGR-1051) and the grant SEV-2011-00067 of the Severo Ochoa Program., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Timely Error Detection for Effective Recovery in Light-Lockstep Automotive Systems

  • Hernandez, Carles
  • Abella Ferrer, Jaume|||0000-0001-7951-4028
Safety-relevant systems in the automotive domain often implement features such as lockstep execution for error detection, and reset and re-execution for error correction. Light-lockstep has already been adopted in some such systems due to its relatively low-implementation cost given that it does not require deep changes into nonlockstep hardware. Instead, as only off-core activities (i.e., data/addresses sent) need to be compared across different cores, light-lockstep designs are lowly intrusive. This approach has been proven sufficient to guarantee functional correctness of the system in the presence of errors in the cores, in particular in relation with certification against safety standards such as ISO26262 in the automotive domain. However, error detection in light-lockstep systems may occur long after the error actually occurs, thus jeopardizing timing guarantees, which are as critical as functional ones in hard real-time systems. In this paper, we analyze the timing behavior of errors due to transient and permanent faults in light-lockstep systems. Our results show that the time elapsed until an error is detected can be inordinately large, especially for permanent faults. Based on this observation and building upon the specific characteristics of light-lockstep systems, we propose lightly verbose (LiVe), a new mechanism to enforce the early detection of errors, due to both transient and permanent faults, thus enabling the computation of tight error detection timing bounds. We also analyze how existing mechanisms for error recovery in multicore systems increase their effectiveness when light-lockstep operates in LiVe mode in the context of mixed-criticality workloads., The research leading to these results has received funding from the ARTEMIS Joint Undertaking VeTeSS project under grant agreement number 295311. This work has also been funded by the Ministry of Science and Technology of Spain under contract TIN2012-34557 and HiPEAC. Jaume Abella has been partially supported by the Ministry of Economy and Competitiveness under Ramon y Cajal postdoctoral fellowship number RYC-2013-14717., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

TM-dietlibc: A TM-aware real-world system library

  • Smiljkovic, Vesna
  • Nowack, Martin
  • Miletic, Nebojša
  • Harris, Tim
  • Unsal, Osman Sabri
  • Cristal Kestelman, Adrián|||0000-0003-1277-9296
  • Valero Cortés, Mateo|||0000-0003-2917-2482
The simplicity of concurrent programming with Transactional Memory (TM) and its recent implementation in mainstream processors greatly motivates researchers and industry to investigate this field and propose new implementations and optimizations. However, there is still no standard C system library which a wide range of TM developers can adopt. TM application developers have been forced to avoid library calls inside of transactions or to execute them irrevocably (i.e. in serial order). In this paper, we present the first TM-aware system library, a complex software implementation integrated with TM principles and suited for software (STM), hardware (HTM) and hybrid TM (HyTM). The library we propose is derived from a modified lock-based implementation and can be used with the existing standard C API. In our work, we describe design challenges and code optimizations that would be specific to any TMbased system library or application. We argue about system call execution within transactions, highlighting the possibility of unexpected results from threads. For this reason we propose: (1) a mechanism for detecting conflicts over kernel data in user space, and (2) a new barrier to allow hybrid TM to be used effectively with system libraries. Our evaluation includes different TM implementations and the focus is on memory management and file operations since they are widely used in applications and require additional mechanisms for concurrent execution. We show the benefit we gain with our libc modifications providing parallel execution as much as possible. The library we propose shows high scalability when linked with STM and HTM. For file operations it shows on average a 1.1, 2.6 and 3.7x performance speedup for 8 cores using HyTM, STM and HTM, respectively (over a lock-based single-threaded execution). For a red-black tree it shows on average 3.14x performance speedup for 8 cores using STM (over a multi-read single-threaded execution)., Peer Reviewed
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Revisiting LP-NUCA Energy Consumption: Cache Access Policies and Adaptive Block Dropping

  • Suárez Gracia, Darío
  • Ferrerón, Alexandra
  • Montesano del Campo, Luis
  • Monreal Arnal, Teresa
  • Viñals Yúfera, Víctor
Cache working-set adaptation is key as embedded systems move to multiprocessor and Simultaneous Multithreaded Architectures (SMT) because interthread pollution harms system performance and battery life. Light-Power NUCA (LP-NUCA) is a working-set adaptive cache that depends on temporal-locality to save energy. This work identifies the sources of energy waste in LP-NUCAs: parallel access to the tag and data arrays of the tiles and low locality phases with useless block migration. To counteract both issues, we prove that switching to serial access reduces energy without harming performance and propose a machine learning Adaptive Drop Rate (ADR) controller that minimizes the amount of replacement and migration when locality is low. This work demonstrates that these techniques efficiently adapt the cache drop and access policies to save energy. They reduce LP-NUCA consumption 22.7% for 1SMT. With interthread cache contention in 2SMT, the savings rise to 29%. Versus a conventional organization, energy--delay improves 20.8% and 25% for 1- and 2SMT benchmarks, and, in 65% of the 2SMT mixes, gains are larger than 20%.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Hardware Architectural Support for Caching Partitioned Reconfigurations in Reconfigurable Systems

  • Clemente, Juan Antonio
  • Gran Tejero, Rubén
  • Chocano, Abel
  • del Prado, Carlos
  • Resano, Javier
The efficiency of the reconfiguration process in modern field-programmable gate arrays (FPGAs) can improve drastically if an on-chip configuration memory is included in the system, because it can reduce both the reconfiguration latency and its energy consumption. However, the FPGA on-chip memory resources are very limited. Thus, it is very important to manage them effectively in order to improve the reconfiguration process as much as possible, even when the size of the on-chip configuration memory is small. This paper presents a hardware implementation of an on-chip configuration memory controller that efficiently manages run-time reconfigurations. In order to optimize the use of the on-chip memory, this controller includes support to deal with configurations that have been divided into blocks of customizable size. When a reconfiguration must be carried out, our controller provides the blocks stored on-chip and looks for the remaining blocks by accessing to the off-chip configuration memory. Moreover, it dynamically decides which blocks must be stored on-chip. To this end, the designed controller implements a simple but efficient technique that allows maximizing the benefits of the on-chip memories. Experimental results will demonstrate that its implementation cost is very affordable and that it introduces negligible run-time management overheads.
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

On the selection of adder unit in energy efficient vector processing

  • Ratkovic, Ivan
  • Palomar Pérez, Óscar
  • Stanic, Milan
  • Unsal, Osman Sabri
  • Cristal Kestelman, Adrián
  • Valero Cortés, Mateo
Vector processors are a very promising solution for mobile devices and servers due to their inherently energy-efficient way of exploiting data-level parallelism. Previous research on vector architectures predominantly focused on performance, so vector processors require a new design space exploration to achieve low power. In this paper, we present a design space exploration of adder unit for vector processors (VA), as it is one of the crucial components in the core design with a non-negligible impact in overall performance and power. For this interrelated circuit-architecture exploration, we developed a novel framework with both architectural- and circuit-level tools. Our framework includes both design- (e.g. adder's family type) and vector architecture-related parameters (e.g. vector length). Finally, we present guidelines on the selection of the most appropriate VA for different types of vector processors according to different sets of metrics of interest. For example, we found that 2-lane configurations are more EDP (Energy×Delay)-efficient than single lane configurations for low-end mobile processors., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Characterizing fault propagation in safety-critical processor designs

  • Espinosa, Jaime
  • Hernandez, Carles
  • Abella Ferrer, Jaume
Achieving reduced time-to-market in modern electronic designs targeting safety critical applications is becoming very challenging, as these designs need to go through a certification step that introduces a non-negligible overhead in the verification and validation process. To cope with this challenge, safety-critical systems industry is demanding new tools and methodologies allowing quick and cost-effective means for robustness verification. Microarchitectural simulators have been widely used to test reliability properties in different domains but their use in the process of robustness verification remains yet to be validated against other accepted methods such as RTL or gate-level simulation. In this paper we perform fault injections in an RTL model of a processor to characterize fault propagation. The results and conclusions of this characterization will serve to devise to what extent fault injection methodologies for robustness verification using microarchitectural simulators can be employed., The research leading to these results has received funding from the ARTEMIS Joint Undertaking VeTeSS project under grant agreement number 295311. This work has also been funded by the Ministry of Economy and Competitiveness of Spain under contract TIN2012-34557 and HiPEAC. Jaume Abella is partially supported by the Ministry of Economy and Competitiveness under Ramon y Cajal postdoctoral fellowship number RYC-2013-14717., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Data placement in HPC architectures with heterogeneous off-chip memory

  • Pavlovic, Milan
  • Puzovic, Nikola
  • Ramírez Bellido, Alejandro
The performance of HPC applications is often bounded by the underlying memory system's performance. The trend of increasing the number of cores on a chip imposes even higher memory bandwidth and capacity requirements. The limitations of traditional memory technologies are pushing research in the direction of hybrid memory systems that, besides DRAM, include one or more modules based on some of the higher-density non-volatile memory technologies, where one of them will provide the required bandwidth, while the other will provide the required capacity for the application. This creates many challenges with data placement and migration policies between the modules of such hybrid memory system. In this paper, we propose an architecture with a hybrid memory design that places two technologically different memory modules in a flat address space. On such system, we evaluate several HPC workloads against different data placement and migration policies, compare their performance by means of execution time and the number of non-volatile memory writes, and consider how it can be applied to the future HPC architectures. Our results show that the hybrid memory system with dynamic page migration and limited DRAM capacity, can achieve performance that is comparable to a hypothetical, hard to implement, DRAM-only system., Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

On the usefulness of object tracking techniques in performance analysis

  • Llort Sánchez, Germán
  • Servat, Harald
  • González García, Juan
  • Giménez Lucas, Judit
  • Labarta Mancho, Jesús José
Understanding the behavior of a parallel application is crucial if we are to tune it to achieve its maximum performance. Yet the behavior the application exhibits may change over time and depend on the actual execution scenario: particular inputs and program settings, the number of processes used, or hardware-specific problems. So beyond the details of a single experiment a far more interesting question arises: how does the application behavior respond to changes in the execution conditions? In this paper, we demonstrate that object tracking concepts from computer vision have huge potential to be applied in the context of performance analysis. We leverage tracking techniques to analyze how the behavior of a parallel application evolves through multiple scenarios where the execution conditions change. This method provides comprehensible insights on the influence of different parameters on the application behavior, enabling us to identify the most relevant code regions and their performance trends. Copyright 2013 ACM., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Aeneas: A tool to enable applications to effectively use non-relational databases

  • Cugnasco, Cesare
  • Hernández, Roger
  • Becerra Fontal, Yolanda
  • Torres Viñals, Jordi
  • Ayguadé Parra, Eduard
Non-relational databases arise as a solution to solve the scalability problems of relational databases when dealing with big data applications. However, they are highly configurable prone to user decisions that can heavily affect their performance. In order to maximize the performance, different data models and queries should be analyzed to choose the best fit. This may involve a wide range of tests and may result in productivity issues. We present Aeneas, a tool to support the design of data management code for applications using non-relational databases. Aeneas provides an easy and fast methodology to support the decision about how to organize and retrieve data in order to improve the performance., Peer Reviewed, Postprint (author’s final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación

FLEX-MPI: An MPI Extension for Supporting Dynamic Load Balancing on Heterogeneous Non-dedicated Systems

  • Martin, Gonzalo
  • Marinescu, Maria-Cristina
  • Singh, David E.
  • Carretero, Jesús
This paper introduces FLEX-MPI, a novel runtime approach for the dynamic load balancing of MPI-based SPMD applications running on heterogeneous platforms in the presence of dynamic external loads. To effectively balance the workload, FLEX-MPI monitors the actual performance of applications via hardware counters and the MPI profiling interface—with a negligible overhead and minimal code modifications. Our results show that by using this approach the execution time of an application may be significantly reduced., This work has been partially supported by the Spanish Ministry of Science under the grant IPT-430000-2010-14., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

TM-dietlibc: A TM-aware real-world system library

  • Smiljkovic, Vesna
  • Nowack, Martin
  • Miletic, Nebojša
  • Harris, Tim
  • Unsal, Osman Sabri
  • Cristal Kestelman, Adrián
  • Valero Cortés, Mateo
The simplicity of concurrent programming with Transactional Memory (TM) and its recent implementation in mainstream processors greatly motivates researchers and industry to investigate this field and propose new implementations and optimizations. However, there is still no standard C system library which a wide range of TM developers can adopt. TM application developers have been forced to avoid library calls inside of transactions or to execute them irrevocably (i.e. in serial order). In this paper, we present the first TM-aware system library, a complex software implementation integrated with TM principles and suited for software (STM), hardware (HTM) and hybrid TM (HyTM). The library we propose is derived from a modified lock-based implementation and can be used with the existing standard C API. In our work, we describe design challenges and code optimizations that would be specific to any TMbased system library or application. We argue about system call execution within transactions, highlighting the possibility of unexpected results from threads. For this reason we propose: (1) a mechanism for detecting conflicts over kernel data in user space, and (2) a new barrier to allow hybrid TM to be used effectively with system libraries. Our evaluation includes different TM implementations and the focus is on memory management and file operations since they are widely used in applications and require additional mechanisms for concurrent execution. We show the benefit we gain with our libc modifications providing parallel execution as much as possible. The library we propose shows high scalability when linked with STM and HTM. For file operations it shows on average a 1.1, 2.6 and 3.7x performance speedup for 8 cores using HyTM, STM and HTM, respectively (over a lock-based single-threaded execution). For a red-black tree it shows on average 3.14x performance speedup for 8 cores using STM (over a multi-read single-threaded execution)., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Parallelizing general histogram application for CUDA architectures

  • Milic, Ugljesa
  • Gelado Fernandez, Isaac
  • Puzovic, Nikola
  • Ramírez Bellido, Alejandro
  • Tomasevic, Milo
Histogramming is a tool commonly used in data analysis. Although its serial version is simple to implement, providing an efficient and scalable way to parallelize it can be challenging. This especially holds in case of platforms that contain one or several massively parallel devices like CUDA-capable GPUs due to issues with domain decomposition, use of global memory and similar. In this paper we compare two approaches for implementing general purpose histogramming on GPUs. The first algorithm is based on private copies of bin counters stored in shared memory for each block of threads. The second one uses the Thrust library to sort the input elements and then to search for upper bounds according to bin widths. For both algorithms we analyze how the speedup over the sequential version depends on the size of input collection, number of bins, and the type and distribution of input elements. We also implement overlapping of data transfers between host CPU and CUDA device with kernel execution. For both algorithms we analyze the pros and cons in detail. For example, privatization strategy can be up to 2x faster than sort-search with realistic inputs, but can only support a limited number of bins. On the other hand, sort-search strategy has about 50% higher speedup than privatization when we use characters as input and can support unlimited number of bins. Finally, we perform an exploration to determine the optimal algorithm depending on the characteristics and values of input parameters., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Parallel algorithm for simulating the spatial transmission of influenza in EpiGraph

  • Martín, Gonzalo
  • Singh, David E.
  • Marinescu, Maria-Cristina
  • Carretero, Jesús
This paper introduces an approach to modeling and simulating the propagation of u-like infectious diseases over large, widely spread urban areas connected by transportation networks. We incorporate geographic location and a transportation model into our region-based, closed-world EpiGraph simulator to realistically model the movement of the virus between different geographic regions. The resulting simulator can assist in understanding how outbreaks propagate between far apart regions due to the movement of people outside their base location. This paper describes the MPI-based implementation of EpiGraph and its performance evaluation when simulating large-scale scenarios. We evaluate the simulator both on a distributed memory system and on a shared memory system., This work has been partially supported by the Spanish Ministry of Science under the grant IPT-430000-2010-14., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Improving performance of all-to-all communication through loop scheduling in PGAS environments

  • Alvanos, Michail
  • Tanase, Gabriel
  • Farreras Esclusa, Montserrat
  • Tiotto, Ettore
  • Amaral, José Nelson
  • Martorell Bofill, Xavier
Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Towards general purpose computations on low-end mobile GPUs

  • Trompouki, Matina M.
  • Kosmidis, Leonidas
GPUs traditionally offer high computational capabilities, frequently higher than their CPU counterparts. While high-end mobile GPUs vendors introduced recently general purpose APIs, such as OpenCL, to leverage their computational power, the vast majority of the mobile devices lack such support. Despite that their graphics APIs have similarities with desktop graphics APIs, they have significant differences, which prevent the use of well-known techniques that offer general-purpose computations over such interfaces. In this paper we show how these obstacles can be overcome, in order to achieve general purpose programmability of these devices. As a proof of concept we implemented our proposal on a real embedded platform (Raspberry Pi) based on Broadcom's VideoCore IV GPU, obtaining a speedup of 7.2× over the CPU., This work has been partially supported by the Spanish Ministry of Science and Innovation under grant TIN2015-65316-P and the HiPEAC Network of Excellence. Leonidas Kosmidis is also funded by the Spanish Ministry of Education under the FPU grant AP2010-4208., Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Dense matrix computations on NUMA architectures with distance-aware work stealing

  • Al-Omairy, Rabab
  • Miranda Álamo, Guillermo
  • Ltaief, Hatem
  • Badia Sala, Rosa Maria
  • Martorell Bofill, Xavier
  • Labarta Mancho, Jesús José
  • Keyes, David E.
We employ the dynamic runtime system OmpSs to decrease the overhead of data motion in the now ubiquitous non-uniform memory access (NUMA) high concurrency environment of multicore processors. The dense numerical linear algebra algorithms of Cholesky factorization and symmetric matrix inversion are employed as representative benchmarks. Work stealing occurs within an innovative NUMA-aware scheduling policy to reduce data movement between NUMA nodes. The overall approach achieves separation of concerns by abstracting the complexity of the hardware from the end users so that high productivity can be achieved. Performance results on a large NUMA system outperform the state-of-the-art existing implementations up to a two fold speedup for the Cholesky factorization, as well as the symmetric matrix inversion, while the OmpSs-enabled code maintains strong similarity to its original sequential version., The authors would like to thank the National Institute for Computational Sciences for granting us access on the Nautilus system. The KAUST authors acknowledge support of the Extreme Computing Research Center. The BSC-affiliated authors thankfully acknowledges the support of the European Commission through the HiPEAC-3 Network of Excellence (FP7-ICT 287759), Intel-BSC Exascale Lab and IBM/BSC Exascale Initiative collaboration, Spanish Ministry of Education (FPU), Computación de Altas Prestaciones VI (TIN2012-34557), Generalitat de Catalunya (2014-SGR-1051) and the grant SEV-2011-00067 of the Severo Ochoa Program., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Timely Error Detection for Effective Recovery in Light-Lockstep Automotive Systems

  • Hernandez, Carles
  • Abella Ferrer, Jaume
Safety-relevant systems in the automotive domain often implement features such as lockstep execution for error detection, and reset and re-execution for error correction. Light-lockstep has already been adopted in some such systems due to its relatively low-implementation cost given that it does not require deep changes into nonlockstep hardware. Instead, as only off-core activities (i.e., data/addresses sent) need to be compared across different cores, light-lockstep designs are lowly intrusive. This approach has been proven sufficient to guarantee functional correctness of the system in the presence of errors in the cores, in particular in relation with certification against safety standards such as ISO26262 in the automotive domain. However, error detection in light-lockstep systems may occur long after the error actually occurs, thus jeopardizing timing guarantees, which are as critical as functional ones in hard real-time systems. In this paper, we analyze the timing behavior of errors due to transient and permanent faults in light-lockstep systems. Our results show that the time elapsed until an error is detected can be inordinately large, especially for permanent faults. Based on this observation and building upon the specific characteristics of light-lockstep systems, we propose lightly verbose (LiVe), a new mechanism to enforce the early detection of errors, due to both transient and permanent faults, thus enabling the computation of tight error detection timing bounds. We also analyze how existing mechanisms for error recovery in multicore systems increase their effectiveness when light-lockstep operates in LiVe mode in the context of mixed-criticality workloads., The research leading to these results has received funding from the ARTEMIS Joint Undertaking VeTeSS project under grant agreement number 295311. This work has also been funded by the Ministry of Science and Technology of Spain under contract TIN2012-34557 and HiPEAC. Jaume Abella has been partially supported by the Ministry of Economy and Competitiveness under Ramon y Cajal postdoctoral fellowship number RYC-2013-14717., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Supercomputing with commodity CPUs: are mobile SoCs ready for HPC?

  • Rajovic, Nikola
  • Carpenter, Paul Matthew
  • Gelado Fernandez, Isaac
  • Puzovic, Nikola
  • Ramírez Bellido, Alejandro
  • Valero Cortés, Mateo
Best Student Paper Award, a la SC13: International Conference for High Performance Computing, Networking, Storage and Analysis, In the late 1990s, powerful economic forces led to the adoption of commodity desktop processors in high-performance computing. This transformation has been so effective that the June 2013 TOP500 list is still dominated by x86. In 2013, the largest commodity market in computing is not PCs or servers, but mobile computing, comprising smart-phones and tablets, most of which are built with ARM-based SoCs. This leads to the suggestion that once mobile SoCs deliver sufficient performance, mobile SoCs can help reduce the cost of HPC. This paper addresses this question in detail. We analyze the trend in mobile SoC performance, comparing it with the similar trend in the 1990s. We also present our experience evaluating performance and efficiency of mobile SoCs, deploying a cluster and evaluating the network and scalability of production applications. In summary, we give a first answer as to whether mobile SoCs are ready for HPC., Peer Reviewed, Award-winning, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

PerfBound: Conserving Energy with Bounded Overheads in On/Off-Based HPC Interconnects

  • Saravanan, Karthikeyan P.
  • Carpenter, Paul Matthew
Energy and power are key challenges in high-performance computing. System energy efficiency must be significantly improved, and this requires greater efficiency in all subcomponents. An important target of optimization is the interconnect, since network links are always on, consuming power even during idle periods. A large number of HPC machines have a primary interconnect based on Ethernet (about 40 percent of TOP500 machines), which, since 2010, has included support for saving power via Energy Efficient Ethernet (EEE). Nevertheless, it is unlikely that HPC interconnects would use these energy saving modes unless the performance overhead is known and small. This paper presents PerfBound, a self-contained technique to manage on/off-based networks such as EEE, minimizing interconnect link energy consumption subject to a bound on the performance degradation. PerfBound does not require changes to the applications and it uses only local information already available at switches and NICs without introducing additional communication messages, and is also compatible with multi-hop networks. PerfBound is evaluated using traces from a production supercomputer. For twelve out of fourteen applications, PerfBound has high energy savings, up to 70 percent for only 1 percent performance degradation. This paper also presents DynamicFastwake, which extends PerfBound to exploit multiple low-power states. DynamicFastwake achieves an energy-delay product 10 percent lower than the original PerfBound technique, This research was supported by European Union’s 7th Framework Programme [FP7/2007-2013] under the Mont-Blanc-3 (FP7-ICT-671697) and EUROSERVER (FP7-ICT-610456) projects, the Ministry of Economy and Competitiveness of Spain (TIN2012-34557 and TIN2015-65316), Generalitat de Catalunya (FI-AGAUR 2012 FI B 00644, 2014-SGR-1051 and 2014-SGR-1272), the European Union’s Horizon2020 research and innovation programme under the HiPEAC-3 Network of Excellence (ICT-287759), and the Severo Ochoa Program (SEV-2011-00067) of the Spanish Government., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Producer-consumer: the programming model for future many-core processors

  • Prat Pérez, Arnau
  • Domínguez Sal, David
  • Larriba Pey, Josep
  • Troncoso, Pedro
The massive addition of cores on a chip is adding more pressure to the accesses to main memory. In order to avoid this bottleneck, we propose the use of a simple producer-consumer model, which allows for the temporary results to be transferred directly from one task to another. These data transfer operations are performed within the chip, using on-chip memory, thus avoiding costly off-chip memory accesses. We implement this model on a real many-core processor, the 48-core Intel Single-chip Cloud Computer processor using its on-chip memory facilities. We find that the Producer-Consumer model adapts to such architectures and allow to achieve good task and data parallelism. For the evaluation of the proposed platform we implement a graph-based application using the Producer- Consumer model. Our tests show that the model scales very well as it takes advantage of the on-chip memory. The execution times of our implementation are up to 9 times faster than the baseline implementation, which relies on storing the temporary results to main memory., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Identifying critical code sections in dataflow programming models

  • Subotic, Vladimir
  • Sancho, Jose Carlos
  • Labarta Mancho, Jesús José
  • Valero Cortés, Mateo
The years of practice in optimizing applications point that the major issue is focus - identifying the critical code section whose optimization would yield the highest overall speedup. While this issue is mainly solved for sequential applications, it remains a serious hurdle in the world of parallel computing. Furthermore, the newest dataflow parallel programming models expose very irregular parallelism, making the identification of the critical code section even harder. To address this issue, we designed an environment that identifies critical code sections in applications. The programmer can use this environment to estimate the potential benefits of the optimization for a specific parallel platform. This is very important because the programmer can anticipate the benefits of his optimization and assure that the optimization is worth the effort. Furthermore, we showed that in many applications, the choice of the critical code section decisively depends on the configuration of the target machine. For instance, in HP Linpack, optimizing a task that takes 0.49% of the total computation time yields the overall speedup of less than 0.25% on one machine, and at the same time, yields the overall speedup of more than 24% on a machine with different number of cores., Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Business process mining from e-commerce web logs

  • Poggi Mastrokalo, Nicolas
  • Muthusamy, Vinod
  • Carrera Pérez, David
  • Khalaf, Rania
The dynamic nature of the Web and its increasing importance as an economic platform create the need of new methods and tools for business efficiency. Current Web analytic tools do not provide the necessary abstracted view of the underlying customer processes and critical paths of site visitor behavior. Such information can offer insights for businesses to react effectively and efficiently. We propose applying Business Process Management (BPM) methodologies to e-commerce Website logs, and present the challenges, results and potential benefits of such an approach. We use the Business Process Insight (BPI) platform, a collaborative process intelligence toolset that implements the discovery of loosely-coupled processes, and includes novel process mining techniques suitable for the Web. Experiments are performed on custom click-stream logs from a large online travel and booking agency. We first compare Web clicks and BPM events, and then present a methodology to classify and transform URLs into events. We evaluate traditional and custom process mining algorithms to extract business models from real-life Web data. The resulting models present an abstracted view of the relation between pages, exit points, and critical paths taken by customers. Such models show important improvements and aid high-level decision making and optimization of e-commerce sites compared to current state-of-art Web analytics., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

OFAR-CM: Efficient Dragonfly networks with simple congestion management

  • García, Marina
  • Vallejo, Enrique
  • Beivide Palacio, Ramon
  • Valero Cortés, Mateo
  • Rodríguez, Germán
Dragonfly networks are appealing topologies for large-scale Data center and HPC networks, that provide high throughput with low diameter and moderate cost. However, they are prone to congestion under certain frequent traffic patterns that saturate specific network links. Adaptive non-minimal routing can be used to avoid such congestion. That kind of routing employs longer paths to circumvent local or global congested links. However, if a distance-based deadlock avoidance mechanism is employed, more Virtual Channels (VCs) are required, what increases design complexity and cost. OFAR (On-the-Fly Adaptive Routing) is a previously proposed routing that decouples VCs from deadlock avoidance, making local and global misrouting affordable. However, the severity of congestion with OFAR is higher, as it relies on an escape sub network with low bisection bandwidth. Additionally, OFAR allows for unlimited misroutings on the escape sub network, leading to unbounded paths in the network and long latencies. In this paper we propose and evaluate OFAR-CM, a variant of OFAR combined with a simple congestion management (CM) mechanism which only relies on local information, specifically the credit count of the output ports in the local router. With simple escape sub networks such as a Hamiltonian ring or a tree, OFAR outperforms former proposals with distance-based deadlock avoidance. Additionally, although long paths are allowed in theory, in practice packets arrive at their destination in a small number of hops. Altogether, OFAR-CM constitutes the first practicable mechanism to the date that supports both local and global misrouting in Dragonfly networks., The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. ERC-2012-Adg-321253- RoMoL, the Spanish Ministry of Science under contracts TIN2010-21291-C02-02, TIN2012-34557, and by the European HiPEAC Network of Excellence. M. García participated in this work while affiliated with the University of Cantabria., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Analysis and RTL correlation of instruction set simulators for automotive microcontroller robustness verification

  • Espinosa, Jaime
  • Hernandez, Carles
  • Abella Ferrer, Jaume
  • de Andres, David
  • Ruiz, Juan C.
Increasingly complex microcontroller designs for safety-relevant automotive systems require the adoption of new methods and tools to enable a cost-effective verification of their robustness. In particular, costs associated to the certification against the ISO26262 safety standard must be kept low for economical reasons. In this context, simulation-based verification using instruction set simulators (ISS) arises as a promising approach to partially cope with the increasing cost of the verification process as it allows taking design decisions in early design stages when modifications can be performed quickly and with low cost. However, it remains to be proven that verification in those stages provides accurate enough information to be used in the context of automotive microcontrollers. In this paper we analyze the existing correlation between fault injection experiments in an RTL microcontroller description and the information available at the ISS to enable accurate ISS-based fault injection., The research leading to these results has received funding from the ARTEMIS Joint Undertaking VeTeSS project under grant agreement number 295311. This work has also been funded by the Ministry of Science and Technology of Spain under contract TIN2012-34557 and HiPEAC. Jaume Abella is partially supported by the Ministry of Economy and Competitiveness under Ramon y Cajal postdoctoral fellowship number RYC-2013-14717., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Thread assignment of multithreaded network applications in multicore/multithreaded processors

  • Radojković, Petar
  • Cakarevic, Vladimir
  • Verdú Mulà, Javier
  • Pajuelo González, Manuel Alejandro
  • Cazorla Almeida, Francisco Javier
  • Nemirovsky, Mario
  • Valero Cortés, Mateo
© 2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., The introduction of multithreaded processors comprised of a large number of cores with many shared resources makes thread scheduling, and in particular optimal assignment of running threads to processor hardware contexts to become one of the most promising ways to improve the system performance. However, finding optimal thread assignments for workloads running in state-of-the-art multicore/multithreaded processors is an NP-complete problem. In this paper, we propose BlackBox scheduler, a systematic method for thread assignment of multithreaded network applications running on multicore/multithreaded processors. The method requires minimum information about the target processor architecture and no data about the hardware requirements of the applications under study. The proposed method is evaluated with an industrial case study for a set of multithreaded network applications running on the UltraSPARC T2 processor. In most of the experiments, the proposed thread assignment method detected the best actual thread assignment in the evaluation sample. The method improved the system performance from 5 to 48 percent with respect to load balancing algorithms used in state-of-the-art OSs, and up to 60 percent with respect to a naive thread assignment., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Applying measurement-based probabilistic timing analysis to buffer resources

  • Kosmidis, Leonidas
  • Vardanega, Tulio
  • Abella Ferrer, Jaume
  • Quiñones, Eduardo
  • Cazorla Almeida, Francisco Javier
The use of complex hardware makes it difficult for current timing analysis techniques to compute trustworthy and tight worst-case execution time (WCET) bounds. Those techniques require detailed knowledge of the internal operation and state of the platform, at both the software and hardware level. Obtaining that information for modern hardware platforms is increasingly difficult. Measurement-Based Probabilistic Timing Analysis (MBPTA) reduces the cost of acquiring the knowledge needed for computing trustworthy and tight WCET bounds. MBPTA based on Extreme Value Theory requires the execution time of processor instructions to be independent and identically distributed (i.i.d.), which can be achieved with some hardware support. Previous proposals show how those properties can be achieved for caches. This paper considers, for the first time, the implications on MBPTA of using buffer resources. Buffers in general, and firstcome first-served (FCFS) buffers in particular, are of paramount importance as the complexity of hardware increases, since they allow managing contention in those resources where multiple requests may be pending. We show how buffers can be used in the context of MBPTA and provide illustrative examples., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Cross-stratum orchestration and flexgrid optical networks for datacenter federations

  • Velasco Esteban, Luis Domingo
  • Asensio Garcia, Adrian
  • Castro Casales, Alberto
  • Berral García, Josep Lluís
  • Carrera Pérez, David
  • López, Víctor
  • Fernández Palacios, Juan Pedro
Current inter-data-center connections are configured as static big fat pipes, which entails large bit rate over-provisioning and thus high operational costs for DC operators. On the other hand, network operators cannot share such connections between customers, because DC traffic varies greatly over time. Those connections are used to perform virtual machine migration and database synchronization among federated DCs, allowing elastic DC operations. To improve resource utilization and save costs, dynamic inter-DC connectivity is currently being targeted from a research point of view and in standardization form. In this article, we show that dynamic connectivity is not enough to guarantee elastic DC operations and might lead to poor performance provided that not enough overprovisioning of network resources is performed. To alleviate it to some extent, we propose using the flexgrid optical technology that enables finer spectrum granularity adaptation and the ability to dynamically increase and decrease the amount of optical resources assigned to connections. DCs can be interconnected through a flexgrid-based network controlled using a centralized software defined network, based on the architecture currently being proposed by the IETF; a cross-stratum orchestrator architecture coordinates DC and network elastically. Illustrative results show that dynamic elastic connectivity provides benefits by reducing the amount of overprovisioned network resources and facilitating elastic DC operations., The research leading to these results has received funding from the European Community’s Seventh Framework Pro- gramme FP7/2007-2013 under grant agreement no. 317999 IDEALIST project. Moreover, it was supported by the Span- ish science ministry through the TEC2011-27310 ELASTIC project., Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

EcoTM: Conflict-aware economical unbounded hardware transactional memory

  • Tomić, Saša
  • Akpinar, Ege
  • Cristal Kestelman, Adrián
  • Unsal, Osman Sabri
  • Valero Cortés, Mateo
Transactional Memory (TM) is a promising paradigm for parallel programming. TM allows a thread to make a series of memory accesses as a single, atomic, transaction, while avoiding deadlocks, livelocks, and other problems commonly associated with lock-based programming. In this paper we explore Hardware support for TM (HTM). In particular, we explore how HTM can efficiently support transactions of nearly unlimited size. For this purpose we propose EcoTM, an economical unbounded HTM that improves the efficiency of conflict detection between very large transactions by activating conflict-detection logic only for potentially-conflicting locations: shared and speculatively modified. EcoTM detects the potentially-conflicting locations automatically, without any program annotations. We evaluate EcoTM performance by comparing it with ideal-lazy HTM, unbounded eager HTM with perfect signatures, and LogTM-SE. Our evaluations show that EcoTM has similar performance as the ideal-lazy HTM, 8.8% better than the eager- perfect HTM, and over 35.7% better than LogTM-SE, on the average., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

HPC system software for regular and irregular parallel applications

  • Morari, Alessandro
  • Valero Cortés, Mateo
The upcoming generation of system software for High Performance Computing is expected to provide a richer set of functionalities without compromising application performance. This Ph.D. thesis addresses the problem of designing scalable system software for both regular and irregular applications. The contributions are two-fold. First, we evaluate the drawbacks of current HPC system software for regular applications. We describe a methodology to precisely measure jitter on a general-purpose OS. Considering a lightweight operating system (IBM CNK), we analyze the overhead of adding support for a missing feature such as dynamic memory management. Second, we focus on irregular applications and build a specialized runtime system to enhance this kind of applications on common HPC flop intensive systems. The proposed runtime system provides a global address space abstraction of a distributed memory machine combined with a transparent fork/join execution model and it also includes lightweight multithreading and network message aggregation., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Circuit design of a novel adaptable and reliable L1 data cache

  • Seyedi, Azam
  • Yalcin, Gulay
  • Unsal, Osman Sabri
  • Cristal Kestelman, Adrián
This paper proposes a novel adaptable and reliable L1 data cache design (Adapcache) with the unique capability of automatically adapting itself for different supply voltage levels and providing the highest reliability. Depending on the supply voltage level, Adapcache defines three operating modes: In high supply voltages, Adapcache provides reliability through single-bit parity. In middle range of supply voltages, Adapcache writes data to two separate cache-lines simultaneously in order to use one line for error recovery when the other line is faulty. In near threshold supply voltages, Adapcache writes data to three separate cache-lines simultaneously in order to provide the correct data based on bitwise majority voter. We design and simulate one embodiment of the Adapcache as a 64-KB L1 data cache with 45-nm CMOS technology at 2GHz processor frequency for almost nominal supply voltages (1V-0.6V), at 900MHz for middle supply voltages (0.6V-0.4V), and at 400MHz for near threshold supply voltages (0.4V-0.32V). According to our experimental results, the energy reduction and latency as well as cache capacity usage are improved compared to typical previous proposals, Triple Modular Redundancy (TMR) and Double Modular Redundancy (DMR) techniques and also to the state of the art proposal, Parichute Error Correction Code (ECC)., Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Improving the energy efficiency of hardware-assisted watchpoint systems

  • Karakostas, Vasileios
  • Tomić, Saša
  • Unsal, Osman Sabri
  • Nemirovsky, Mario
  • Cristal Kestelman, Adrián
Hardware-assisted watchpoint systems enhance the execution of numerous dynamic software techniques, such as memory protection, module isolation, deterministic execution, and data race detection. In this paper, we show that previous hardware proposals may introduce significant energy overheads, and propose WatchPoint Filtering (WPF), a novel filtering mechanism that eliminates unnecessary watchpoint checks. We evaluate WPF on two state-of-the-art proposals for hardware-assisted watchpoints using two common memory checkers. WPF eliminates 83% of the watchpoint checks (up to 99.7%) and reduces 57% of the dynamic energy overhead (up to 78%) on average, without introducing additional performance execution overhead., Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

An OpenMP* barrier using SIMD instructions for Intel® Xeon Phi™ coprocessor

  • Caballero, Diego
  • Duran González, Alejandro
  • Martorell Bofill, Xavier
Barrier synchronisation is a widely-studied topic since the supercomputer era due to its significant impact on the overall performance of parallel applications. With the current shift to many-core architectures, such as the Intel® Many Integrated Core Architecture, software barriers need to be revisited from an on-chip point of view to exploit their new specific resources. In this paper, we propose a tree-based barrier that takes advantage of SIMD instructions and the inter-thread cache locality provided by the 4-way SMT of the Intel® Xeon PhiTM coprocessor. Our SIMD approach shows a speed-up of up to 2.84x over the default Intel OpenMP* barrier in the EPCC barrier microbenchmark. It also improves by up to 60% and 21% the Livermore Loop kernel number six and the NAS MG benchmark, respectively., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Tessellation: Refactoring the OS around explicit resource containers with continuous adaptation

  • Colmenares, Juan A.
  • Eads, Gage
  • Hofmeyry, Steven
  • Bird, Sarah L.
  • Moretó Planas, Miquel
  • Chou, David
  • Gluzman, Brian
  • Roman, Eric
  • Bartolini, Davide B.
  • Mor, Nitesh
  • Asanovic, Krste
  • Kubiatowicz, John D.
Adaptive Resource-Centric Computing (ARCC) enables a simultaneous mix of high-throughput parallel, real-time, and interactive applications through automatic discovery of the correct mix of resource assignments necessary to achieve application requirements. This approach, embodied in the Tessellation manycore operating system, distributes resources to QoS domains called cells. Tessellation separates global decisions about the allocation of resources to cells from application-specific scheduling of resources within cells. We examine the implementation of ARCC in the Tessellation OS, highlight Tessellation's ability to provide predictable performance, and investigate the performance of Tessellation services within cells., This research is supported by Microsoft (Award #024263), Intel (Award #024894), matching U.C. Discovery funding (Award #DIG07-102270), and DOE ASCR FastOS Grant #DE-FG02-08ER25849. Additional support comes from Par Lab aliates National Instruments, Nokia, NVIDIA, Oracle, and Samsung. No part of this paper repre- sents views and opinions of the sponsors mentioned above. We thank other Par Lab members for their collaboration and feedback. J. A. Colmenares participated in this work while he was a post-doctoral scholar at UC Berkeley. M. Moreto is supported by a MEC/Fulbright Fellowship., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Framework for a productive performance optimization

  • Servat, Harald
  • Llort, German
  • Huck, Kevin A.
  • Giménez Lucas, Judit
  • Labarta Mancho, Jesús José
Modern supercomputers deliver large computational power, but it is difficult for an application to exploit such power. One factor that limits the application performance is the single node performance. While many performance tools use the microprocessor performance counters to provide insights on serial node performance issues, the complex semantics of these counters pose an obstacle to an inexperienced developer. We present a framework that allows easy identification and qualification of serial node performance bottlenecks in parallel applications. The output of the framework is precise and it is capable of correlating performance inefficiencies with small regions of code within the application. The framework not only points to regions of code but also simplifies the semantics of the performance counters into metrics that refer to processor functional units. With such information the developer can focus on the identified code and improve it by knowing which processor execution unit is degrading the performance. To demonstrate the usefulness of the framework we apply it to three already optimized applications using realistic inputs and, according to the results, modify their source code. By doing modifications that require little effort, we successfully increase the applications’ performance from 10% to 30% and thus shorten the time required to reach the solution and/or allow facing increased problem sizes., Peer Reviewed, Preprint
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Enabling distributed key-value stores with low latency-impact snapshot support

  • Polo Bardés, Jordà
  • Becerra Fontal, Yolanda
  • Carrera Pérez, David
  • Torres Viñals, Jordi
  • Ayguadé Parra, Eduard
  • Spreitzer, Mike
  • Steinder, Malgorzata
Current distributed key-value stores generally provide greater scalability at the expense of weaker consistency and isolation. However, additional isolation support is becoming increasingly important in the environments in which these stores are deployed, where different kinds of applications with different needs are executed, from transactional workloads to data analytics. While fully-fledged ACID support may not be feasible, it is still possible to take advantage of the design of these data stores, which often include the notion of multiversion concurrency control, to enable them with additional features at a much lower performance cost and maintaining its scalability and availability. In this paper we explore the effects that additional consistency guarantees and isolation capabilities may have on a state of the art key-value store: Apache Cassandra. We propose and implement a new multiversioned isolation level that provides stronger guarantees without compromising Cassandra's scalability and availability. As shown in our experiments, our version of Cassandra allows Snapshot Isolation-like transactions, preserving the overall performance and scalability of the system., This work is partially supported by the Ministry of Science and Technology of Spain and the European Union’s FEDER funds (TIN2007-60625, TIN2012-34557), by the Generalitat de Catalunya (2009-SGR-980), by the BSC-CNS Severo Ochoa program (SEV-2011-00067), by the HiPEAC European Network of Excellence (IST- 004408, FP7-ICT-217068, FP7-ICT-287759), and by IBM through the 2008 and 2010 IBM Faculty Award program., Peer Reviewed, Postprint (author’s final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

A proposal for task-generating loops in OpenMP

  • Teruel, Xavier
  • Klemm, Michael
  • Li, Kelvin
  • Martorell Bofill, Xavier
  • Olivier, Stephen
  • Terboven, Christian
With the addition of the OpenMP* tasking model, programmers are able to improve and extend the parallelization opportunities of their codes. Programmers can also distribute the creation of tasks using a worksharing construct, which allows the generation of work to be parallelized. However, while it is possible to create tasks inside worksharing constructs, it is not possible to distribute work when not all threads reach the same worksharing construct. We propose a new worksharing-like construct that removes this restriction: the taskloop construct. With this new construct, we can distribute work when executing in the context of an explicit task, a single, or a master construct, enabling us to explore new parallelization opportunities in our applications. Although we focus our current work on evaluating expressiveness rather than performance evaluation, we present some initial performance results using a naive implementation for the new taskloop construct based on a lazy task instantiation mechanism., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Improving communication in PGAS environments: Static and dynamic coalescing in UPC

  • Alvanos, Michail
  • Farreras Esclusa, Montserrat
  • Tiotto, Ettore
  • Amaral, José Nelson
  • Martorell Bofill, Xavier
The goal of Partitioned Global Address Space (PGAS) languages is to improve programmer productivity in large scale parallel machines. However, PGAS programs may have many fine-grained shared accesses that lead to performance degradation. Manual code transformations or compiler optimizations are required to improve the performance of programs with fine-grained accesses. The downside of manual code transformations is the increased program complexity that hinders programmer productivity. On the other hand, most compiler optimizations of fine-grain accesses require knowledge of physical data mapping and the use of parallel loop constructs. This paper presents an optimization for the Unified Parallel C language that combines compile time (static) and runtime (dynamic) coalescing of shared data, without the knowledge of physical data mapping. Larger messages increase the network efficiency and static coalescing decreases the overhead of library calls. The performance evaluation uses two microbenchmarks and three benchmarks to obtain scaling and absolute performance numbers on up to 32768 cores of a Power 775 machine. Our results show that the compiler transformation results in speedups from 1.15X up to 21X compared with the baseline versions and that they achieve up to 63% the performance of the MPI versions., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Efficient routing mechanisms for Dragonfly networks

  • García, Marina
  • Vallejo, Enrique
  • Beivide Palacio, Julio Ramón
  • Odriozola, Miguel
  • Valero Cortés, Mateo
High-radix hierarchical networks are cost-effective topologies for large scale computers. In such networks, routers are organized in super nodes, with local and global interconnections. These networks, known as Dragonflies, outperform traditional topologies such as multi-trees or tori, in cost and scalability. However, depending on the traffic pattern, network congestion can lead to degraded performance. Misrouting (non-minimal routing) can be employed to avoid saturated global or local links. Nevertheless, with the current deadlock avoidance mechanisms used for these networks, supporting misrouting implies routers with a larger number of virtual channels. This exacerbates the buffer memory requirements that constitute one of the main constraints in high-radix switches. In this paper we introduce two novel deadlock-free routing mechanisms for Dragonfly networks that support on-the-fly adaptive routing. Using these schemes both global and local misrouting are allowed employing the same number of virtual channels as in previous proposals. Opportunistic Local Misrouting obtains the best performance by providing the highest routing freedom, and relying on a deadlock-free escape path to the destination for every packet. However, it requires Virtual Cut-Through flow-control. By contrast, Restricted Local Misrouting prevents the appearance of cycles thanks to a restriction of the possible routes within super nodes. This makes this mechanism suitable for both Virtual Cut-Through and Wormhole networks. Evaluations show that the proposed deadlock-free routing mechanisms prevent the most frequent pathological issues of Dragonfly networks. As a result, they provide higher performance than previous schemes, while requiring the same area devoted to router buffers., This work has been supported by the Spanish Ministry of Science under contracts TIN2010-21291-C02-02, TIN2012-34557, and by the European HiPEAC Network of Excellence. The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. ERC-2012-Adg-321253-RoMoL. M. Garc´ıa and M. Odriozola participated in this research work while they were affiliated with the University of Cantabria., Peer Reviewed, Postprint (author's final draft)